However, regardless of the definition of consciousness, the answer to the question is actually quite simple. LLMs are NOT conscious, and the reason is that they don’t perform the operations that result in consciousness in animals and people.

Consciousness is the process that animals and people use to analyze complex, real-time, sensory input from their environment; focus attention on specific objects and events; extract additional information from the environment; compare those objects and events with past memories, and decide on a course of action. To put the scale of our sensory environment into perspective, we have 100 million neurons in the eyes, 15 million in the nose, and 10 million sensory neurons in the rest of the body. These senses are connected to over 100 billion neurons in the brain that dynamically analyse this input, react to it, and adapt to it by dynamically rewiring neural connections and storing new memories.

LLM’s don’t have a complex sensory environment and they don’t perform any of the operations related to consciousness. Their entire sensory environment consists of an input prompt. This is fed into their neural network, which is a static collection of weights and biases that was learned during their extensive training process. As a result of this, they generate output text that is presented to the person who typed the prompt. The initial prompt and answer are kept in the context and the LLM waits for the next prompt. To preserve information between sessions, this context is often saved in the user’s profile and reloaded with information about the user the next time they use the system. Some LLMs have a hard-wired “thinking” process that simply feeds variations of the original prompt and special instructions through the neural network multiple times and combines the results. Some have access to additional input from the internet, depending on the prompt.

Let’s take a look at what is involved in routine conscious experience and see if the LLMs have the mechanisms for any of it. For example, as I walk down a street various sights, sounds, and smells are subconsciously analyzed by the brain. The act of walking has been delegated to a subroutine in my brain that takes care of the details. I may be deep in thought about what I should have for lunch and only vaguely aware of what’s happening around me on the street. A person approaches; my brain subconsciously detects this and takes evasive action. I hardly notice. Occasionally the automatic evasive action doesn’t work, my brain brings it to my attention, I focus on the oncoming person, make eye contact, try to figure out which way they are going, move the other way, and avoid a collision. Facial recognition circuitry in my brain recognizes the face of a neighbour approaching and brings it to my attention. I turn my head and focus on the person walking towards me. Sure enough, it’s my neighbour. I say “Hi there”, smile, node my head at them, and move on. I hear a loud honking sound, like a car’s horn. I turn to the direction of the sound and confirm that a car is honking at another pedestrian, not me, no action required, I keep walking. My brain recognizes that I’m approaching an intersection and brings this to my attention. I focus on the traffic light. It is red so I stop and wait.

In all of these cases, we’re not simply taking an input and generating an output like LLMs do. It’s the process of attention, focus, analysis, decision, and action that make up the conscious experience. It’s the process that brains use to survive in complex environments. LLMs don’t use this process and are therefore not conscious. I don’t know how this “attention” thing actually works, but it does. When the brain notices some noteworthy event, it somehow starts the chain of events (focus, analysis, decision, action), that I experience as consciousness.

So why do we think AIs are conscious? It’s simply because they can talk, and that makes them sort of human, and if they are sort of human, they must be conscious.

This does not mean that computers will never be conscious. If you programmed a computer to simulate the workings of a brain and gave it a complex environment to monitor and manipulate, it could potentially experience consciousness. We might not be able to determine if it is actually conscious, or just a Philosophical Zombie, but at least it would be going through the processes that lead to consciousness in animals and people.

As a result, the Toronto City Council voted in 2021 to rename Dundas Street, Dundas Square, and the Dundas subway station. In December 2023 the City Council approved the name “Sankofa Square”. Sankofa is a Ghanaian term that means “reclaiming past teachings that enable us to move forward together”. In 2025 the city officially renamed the square, but decided it would be too expensive to rename the entire street which is 23km long. The cost for renaming the square was $306,000. To rename the street would have cost about $12.7 million. They renamed the nearby Dundas subway station “TMU Station” after Toronto Municipal University. Apparently, the university picked up the $1.7 million dollar cost for that rename.

Ironically, Toronto Municipal University was renamed from Ryerson University in April 2022, when it came to light that it was named after Egerton Ryerson, who was one of the architects of Canada’s residential school system, that separated 150,000 First Nations, Inuit, and Métis children from their families to forcibly assimilate them into Euro-Canadian culture. So, the subway name is actually the result of a double renaming.

Henry Dundas

So, what did Henry Dundas do to deserve having the square renamed. He was a Scottish politician who was the member of parliament for Edinburgh from 1790–1802. He also served as First Lord of the Admiralty, Secretary of State, Home Secretary, and various other positions in the British government. His friend, John Simcoe, the first governor of Upper Canada, name a bunch of stuff after Dundas. With respect to the slave trade, he was an abolitionist, but he was also a practical politician.

In 1791, William Wilberforce introduced a bill in the House of Commons to abolish the trans-Atlantic slave trade. It was soundly defeated by a vote of 163 to 88. In April 1792 he tried again. Dundas supported the bill and presented a petition from Edinburgh residents who supported abolition. However, he realized that the bill was unlikely to pass as it was written and introduced an amendment that asked for a “gradual” phase out rather than an “immediate” abolition. The reason he gave for making the change gradual, was to allow the economies of the West Indies to adapt to the changes. The amendment was adopted, and the bill passed with a vote of 230 to 85, and for the first time, the House of Commons voted to end the slave trade. Dundas then tabled a plan to gradually phase out the slave trade by 1799. Unfortunately, House of Lords rejected the bill and the plan and it was not passed into law, despite Dundas’s efforts. In 1807 the House of Lords finally voted in favour of abolishing trading of slaves, but ownership of slaves remained legal in parts of the British Empire until 1833.

At the time, Dundas was regarded as an abolitionist and it was clear that the addition of the term “gradual” was an honest attempt to get the law through Parliament. It wasn’t Dundas who caused the delay, it was the House of Lords. However, in the 20th century, some historians reopened the case and decided that his action to include the word “gradual” in the bill did in fact delay abolition. The essay, “The Great Delayer” by the historian S. Mullen from Glasgow University, summarizes the claims that Dundas was in cahoots with the West Indies faction to delay the abolition indefinitely. He makes a pretty good case, particularly that Dundas was worried that an immediate abolition would destroy the economies of the West Indies. These ideas were criticized by other historians and the issue is still being debated. In Scotland, they have added the following wording to his monument in Edinburgh…

“While Home Secretary in 1792 and first Secretary of State for War in 1796 he was instrumental in deferring the abolition of the Atlantic slave trade. Slave trading by British ships was not abolished until 1807. As a result of this delay, more than half a million enslaved Africans crossed the Atlantic.”

In Toronto, they renamed the square and the subway station.

Egerton Ryerson

Egerton Ryerson’s case is a bit more clear-cut. He was and educator and Methodist minister who was the Supervisor of the Board on Education in Upper Canada (Ontario) in the mid 1800s. His efforts for school reform resulted in the Common School Act of 1850 that mandated free primary and secondary education for all children, funded by provincial and municipal governments, the formation of school boards, and standardized curriculum and text books. In 1871, under his guidance, education became free and compulsory for students up to age 12. He basically created the modern Ontario school system.

In 1847, George Vardon (Superintendent of Indian Affairs), asked for his advice on “the best method of establishing and conducting Industrial Schools for the benefit of the Aboriginal Indian Tribes”. In the mid-1800s, industrial schools were “residential or day institutions established to care for, train, and reform destitute, vagrant, or neglected children and juvenile delinquents by combining basic education with manual labor and trade skills”. There were many examples already in existence in Europe and the US. So, it was in this context that Ryerson replied with a five-page letter.

In the letter, he focuses on teaching the skills required to become farmers. “It would be a gratifying result to see graduates of our Indian industrial schools become overseers of some the largest farms in Canada”. He emphasized that religious training be a key component and that academics, including reading, writing, mathematics, history, chemistry, biology, geography, music, and art should be included in the curriculum. He recommended that students be paid a small amount and trained in book keeping to manage their income. No mention is made of what the girls should be taught.

The schools were envisioned as working farms where the students provided the labor. They would rise early in the morning, tidy the house, say prayers, have lessons, and then breakfast. They would then work in the fields for 8 to 12 hours depending on the season, with an hour break for lunch. In the evening there would dinner, followed by more lessons, and then bed. From a modern perspective this sounds pretty horrible, more like a forced labour camp than a school.

A large part of the document covers the management of the schools and how they would be governed. The federal government would set the general goals, but the day-to-day management would be left up to the school supervisor, who would be held accountable for the operation of the school. It was clear that he recommended religious groups supervise the schools. He did not recommend forcing Indigenous children to attend the schools, nor punishing them for speaking their Indigenous languages, conditions later introduced by the government or school administrators.

Ryerson died in 1882, 35 years after his infamous letter. Most of the residential school system was set up after his death, but his 1847 letter was certainly influential in ensuring the schools were residential, involved labour, and were run by religious groups. He certainly didn’t anticipate that children would be forcibly removed from their homes to attend these schools.

Conclusion

From my perspective, it looks like Dundas was legitimately trying to navigate a complex political quagmire, weighing the desires of the abolitionists against the economies of the West Indies. His plan to phase out the slave trade by 1799 was blocked by the House of Lords, so blaming the delay on him is somewhat unfair.

Ryerson, certainly wrote the influential five-page recommendation on the curriculum and organization of Residential Schools, and for that must shoulder his share of the blame, but aside from that document, he had no other involvement. In his day, he was regarded and a progressive supporter of education. He also held the common racist views of his times with respect to the indigenous people.

So should we rename streets, squares, subway stations, and universities, simply because the named person has a blot on their record. I guess it depends on how significant the “blot” is, and that is often a matter of opinion. In the Dundas Square case, they had a petition, which is a good way to find out how people feel, and 14,000 people wanted to rename it. We still have Dundas County, the town of Dundas, and Dundas Street (the Toronto section alone is 23km long).

References:

1. https://en.wikipedia.org/wiki/Henry_Dundas,_1st_Viscount_Melville

2. https://en.wikipedia.org/wiki/Sankofa_Square

3. https://historyofparliament.com/wp-content/uploads/2021/09/mullen-henry-dundas-a-e28098great-delayer-of-the-abolition-of-the-transatlantic-slave-trade.pdf

4. https://en.wikipedia.org/wiki/Egerton_Ryerson

5. https://upload.wikimedia.org/wikipedia/commons/3/35/Egerton_Ryerson_on_Residential_Schools.pdf

In the 1970’s we all imagined that it wouldn’t be long until we had flying cars, ate pills for food, had robot servants, used laser guns, learned via telepathy, mined the asteroid belt, and vacationed on the moon. Instead, we got laptops, the Internet, mobile phones, streaming video, chatbots, AncestryDNA, online shopping, and Donald Trump.

On the apocalyptic side, we worried about mass starvation, a coming ice age, destruction of the rain forests, mass extinction, nuclear war, running out of oil, acid rain, and the destruction of the ozone layer. We narrowly avoided nuclear war back in the Cuban Crisis, and did take steps to prevent acid rain and the save the ozone layer. Fortunately, none of the others actually happened. Instead, we got global warming and COVID.

So here we are in 2026. What’s coming next is usually what’s already here, but not yet developed. Here’s some predictions, starting with the fairly obvious ones.

This AI thing will definitely take off. It will revolutionize white collar work. Everyone will have an AI-based personal assistant that does most of their writing for them. We will have LLM based AIs in education, medicine, law, finance, travel, software, design, analysis, marketing, customer support, logistics, services, and many more. Any task that involved reading, writing, listening, or talking will be assisted by LLMs. This use of AI will substantially change these jobs and perhaps eliminate some of them.

AI will also be combined with a variety of walking/talking robots, ranging from the robot servants we envisioned back in the 1970s, to robots that handle specialized and perhaps dangerous jobs. This technology has recently advanced considerably, as was demonstrated at the recent Robot Games in China.

Genetic engineering will be used in medicine to cure a variety of cancers and genetic diseases. Genetically modified plants that are high yield and disease resistant will become common. The same will apply to domestic animals. Genetically modifying humans for non-medical reasons will continue to be taboo, but someone will experiment anyway with creating smarter, stronger, healthier, and more beautiful children. This may be even be legal in certain countries and it may be available on the black market in countries where it is banned. Sports will of course have to contend with genetically modified athletes. Finally, it is inevitable that someone will try to build a dinosaur.

With all the research going on these days, it is inevitable that some major development will occur in the realm of climate change. I have no idea what it will be, but a possible candidate will be a practical way to remove CO2 from the atmosphere. Another possibility is that we finally stop burning gasoline in cars.

Physics is due for another revolution. I predict that the problems introduced by quantum gravity, dark matter and dark energy will be solved and we will finally realize that the Big Bang theory is wrong and that the universe is much older than 14 billion years.

I think the combination of self-driving electric vehicles, UBER, and working from home will revolutionize personal transportation. I look forward to the day when I no longer need a car. When I have to go somewhere a self-driving UBER will pick me up and take me there. Perhaps the proliferation of drones, combined with AI, will lead to actual flying cars.

Space travel is hideously expensive and there’s not really much point to it, so I don’t think we will be vacationing on the moon or mining the asteroid belt any time soon. We’ll focus more on space telescopes and AI controlled probes.

The declining birthrate trend will continue and eventually the world population will stabilize at around 10 billion. However this will lead to the proportion of older adults increasing, which will put pressure on healthcare systems and social security programs. Robots and AI could potentially compensate for the reduced number of young workers that are available.

One major variable is of course world political stability. Who would have thought that the most heavily armed and powerful country in the world could be ruled by someone like Donald Trump, a narcisistic megalomaniac. We’re definately in for more political instability with a widening gap between left and right.

On the apocalypse side, we have a couple serious contenders. Global warming is certainly a real threat, but people are beginning to take it seriously, and perhaps we may have time put the brakes on it by reducing emissions. The other big one is the pending technological singularity where AI takes over and wipes us out. I have dealt with both topics in more detail in…

· Global Warming — Human Impact

· Will AI Kill Us All

On the other hand, something quite unexpected could happen. The Earth could be hit by a comet; we could annihilate ourselves with a nuclear war; or a plague could wipe us out. On the plus side: we could tame nuclear fusion; solve the world population problem; cure cancer; and finally figure out why 42 is the answer to the question of life the universe and everything.

So, what do you think. What will the future hold in the next 50 years.

Many of the early proponents of this idea were quite self-serving. It was first proposed in 1895 by George Hudson, an entomologist from New Zealand. He wanted the extra hours of evening daylight in the summer so he could collect insects in his spare time, after work. He actually presented a paper to the Wellington Philosophical Society proposing a two-hour daylight-saving shift. In England, the idea was first proposed by William Willett, who published a paper on the idea in 1909 and worked hard to push it through Parliament. He was an avid golfer who disliked cutting short his round at dusk. So, daylight savings was first proposed to allow time for golf and bug hunting after work in the summer.

The first place to implement daylight savings was the city of Thunder Bay, Ontario in 1908. The city is on the western side of Lake Superior at the extreme west of a huge Canadian time zone encompassing Quebec and Ontario. All the US States to the south of Thunder Bay are on central time, so at least for the summer, Thunder Bay was in the same time zone as its neighbours. The first countries to implement daylight savings time were Germany and Austria during the first world war in 1916, to conserve coal. Britain and the allies followed suite the next year and the lunacy spread. How advancing the clock saves coal is not completely clear. The idea is that because daylight continues until late in the evening, people won’t have to turn on their lights, thus saving coal. However, light usage is at a minimum in the summer anyway, so there’s not much to save. Many countries bought into this “energy saving” idea and adopted daylight savings during war time. Most of Europe and North America have continued to use it since WW1. There’s a myth in the US that daylight savings was implemented to help the farmers. It wasn’t, and they really don’t like it, because it messes up their cows.

So, what does it do? Let’s take someone living at a latitude of 50 degrees north. I picked this because at 50 degrees north, there are roughly 8 hours of daylight at mid-winter and 16 hours of daylight at the summer solstice, so it makes the math easy. It’s also the average latitude of Europe and close to the Canada/US border. If we use regular time, this means that in mid-winter the sun rises at 8:00am and sets at 4:00pm. In midsummer it would rise at 4:00am and set at 8:00pm. With daylight savings, in midsummer sun rises at 5:00am and sets at 9:00pm, giving us a glorious extra hour of daylight in the evening. But does it… of course not, we just pretend it does.

There are considerable problems caused by changing the clocks. The biggest one is the impact on children. Parents struggle to get their kids onto a regular sleeping schedule and we mess it up with the clock change. Let’s say your kid starts school at 8:30am, so you get them up at 7:30, feed them breakfast and get them off to school. Then March roles around and we set the clock forward one hour and everyone has to get up an hour early. Yesterday’s 6:30 is now 7:30 on the clock, and when the alarm goes off, our bodies still think it’s 6:30. This totally messes up the kids. Then at night we want to send them to bed an hour early. We send them to bed at 9:00pm, but their bodies think it’s 8:00pm and they can’t sleep. All this so some idiot can finish his golf game.

Adult of course suffer the same issues with sleep. A number of sources point to issues with its impact on health and traffic accidents. The Wikipedia article [1] says:

There are measurable adverse effects of clock-shifts on human health.[129] It has been shown to disrupt human circadian rhythms,[130] negatively affecting human health in the process,[131] and that the yearly DST clock-shifts can increase health risks such as heart attacks[132] and traffic accidents.[133][134]

Another problematic area is how computers handle daylight savings time. Most modern computing devices: laptops, phones, & servers, now change from normal time to daylight time automatically, but it has caused significant problems in the past. The consequence of this is that anytime some jurisdiction changes its rules about daylight savings time, it requires an update to the computer system’s time zone data. Also, systems that don’t change automatically, have to be done manually, like the all clocks in my house and the one in my car.

Many people now realize that this daylight savings time is a bit of a boondoggle and some jurisdictions are doing away with it. In Canada, we’ve decided to make it really confusing. The four western provinces: British Columbia, Alberta, Saskatchewan, and Manitoba; and the Yukon Territory, will not move their clocks back this fall and remain on permanent daylight savings time. Ontario, Quebec, and the Maritimes are sticking with the old system and moving them back. Newfoundland has a weird time zone that’s half an hour earlier that the rest of the Maritimes. So at least the west doesn’t have to put up with the clock changes any more, but it’s now more confusing than ever.

Many jurisdictions have chosen permanent daylight savings when they get rid of the time change, but that’s not a good option for countries in the north. For example, if we went with permanent daylight savings time in Glasgow where I was born, it would still be dark at 10:00am in the morning in late December, an it’s only 56 degrees north. So, things will get complicated as different locations opt for permanent regular or daylight time.

It’s amazing that all this nonsense was created because a couple of guys wanted to collect bugs and play golf after work in the summer. Wouldn’t it have been easier just to leave the clocks alone and leave work earlier.

References:

[1] — Daylight Savings Time : Wikipedia

On September 10th 2026, Anthropic issues a 154 page Threat Report that details about 40 cases from between Dec 2025 and Aug 2026, where “threat actors” tried to use Claude for malicious activity. While the majority of cases simply leverage Claude’s ability to generate content and write code, some of the cases of abuse specifically focus on Claude’s expertise in hacking, threat analysis, reverse engineering, vulnerability discovery, exploit generation. Many of them are also combined with credential theft to gain illicit access and then use of automation to amplify the impact of the malicious activity.

Most of the reported incidents involve state-sponsored groups, financially motivated criminals, state propaganda institutions, and politically motivated individuals. One wonders why Anthropic, which is apparently an “AI safety and research company”, is hosting features that can be utilized in such a malicious fashion. The report covers seven categories of abuse. Each of these is discussed below.

Cyber Threats

These are cases where the threat actors use Claude’s cyber-security capabilities to break into computer systems, launch phishing attacks, harvest access credential, and launch cyber attacks at adversaries. They seem quite proud of Claude’s ability to do this and specifically call out the “uplift” that threat actors can gain from using Claude.

“The report also attempts to measure uplift, a term we use to describe the AI capability boost, or how much more harm was caused with AI versus without AI… Many commentators focus on the risk of AI developing exploits at scale. While this is a danger, the risk from AI adoption is more pronounced across the cyber kill chain, where adversaries can operate faster, across a broader and deeper surface area, with fewer resources.”

So, Anthropic is helping hackers to work faster, broader, and deeper, with fewer resources. Despite all the talk, nowhere does the report actually measure uplift.

The incidents in this section include:

· A Russian operative who used customized AI-driven workflows to automated development, infrastructure acquisition, phishing, persistence through command and control, to data exfiltration in attacks on Ukraine and Europe.

· A group called ShinyHunters that use Claude to accelerate their ability to rapidly scan, exploit, and take over target systems.

· Chinese graduate student in Hunan that used Claude as the engineering and orchestration layer for: intrusion attempts against production systems; reconnaissance of foreign-government networks across the Middle East, Europe, and Southeast Asia; a standing vulnerability-research and exploit development effort against major endpoint-security products; malware development; and an intelligence collection platform.

· French-speaking hacktivist used Claude to target European political parties, media, think-tanks, and the SaaS providers used by these organizations. The actor used Claude to develop and debug the exploit that succeeded against at least four victim websites.

In many of these cases, Claude is providing the expertise to assist the attacker. Why does Anthropic allow it to do this.

Influence

These are cases where Claude was used in operations to influence specific factions of the population, usually for political purposes. This typically involved flooding social media with posts that support a specific narrative. The AI can be used to generate content and automate the process. AI agents can actively participate. It often involves creating thousands of fake accounts spewing AI generated slop.

“The actors behind these operations included governments, state-aligned propaganda institutions and state media, as well as private firms selling influence to paying clients, domestic political operators, and in one case an opposition movement in exile.”

I won’t list the cases in detail. Most simply used Claude for some sort of automated content generation. In some cases, this content was directly sent to thousands of spoofed social media accounts. In others, it was distributed via state run media outlets. One case involved an “influence-as-a-service” operation where you can pay for a specific message to be disseminated, but most were state operated and designed to influence voters and public opinion. They involved Russia, Iran, Malaya, Bangladesh, Kenya, but none from Europe or the Americas.

Surveillance

Anthropic reported the AI is being used in three ways for surveillance. The first involves using AI to develop software platforms to support surveillance, and point incidents where a Malian operative used Claude to build a platform to monitor the country’s mobile operators, an Iranian used Claude to build a Firefox plugin that spied on users, and the Chinese built an agent to monitor religious affairs. Using Claude to develop code is one of its primary uses. I don’t see how Anthropic can act as the judge as to what’s good or bad.

The second is to use AI to analyze the collected data. Here they point to several cases where Claude was used to analyze social media posts and create reports on individuals. The third use of AI in surveillance is when it is fully integrated with the state surveillance infrastructure. In other word government surveillance operatives routinely use AI in their jobs.

Conventional Weapons

This section focuses on the use of Claude to develop software for conventional weapons, including firearms, missiles, armed drones, bombs, and other munitions, as well as the targeting and control systems that operate them. The Anthropic threat intelligence team has disrupted the activities on a number of threat actors who are using Claude to develop weapons related software. The report shares details on six cases: three in China, two in Russia, and one in Yemen.

These include the development of missile guidance software in Yemen; a Chinese use of Claude to develop specifications for anti-torpedo weapons; a Russian attempt to develop software to control military drones; Chinese software to jam radar and communications; Russian use of Claude in military procurement; and finally, the Chinese use of Claude to collect intelligence on energy weapons.

Biological Misuse

Anthropic regards this as one of the most serious risks, but they do seem acknowledge that current models may not yet be up to the task. They say…

“To our knowledge, no private company, AI or otherwise, has yet shared evidence of the potential misuse of their platforms for biological weapons development publicly.”

They then go on to presents five cases where the use of their models “could” support the development of biological weapons. These included:

· A request for Claude’s assistance in authoring a grant application for scientific funding for work involving gain-of-function research on the chikungunya virus.

· A researcher using Claude in their research on highly-pathogenic avian influenza (“bird flu”). The research focused on viruses’ adaptation to mammals, and the mechanism by which it causes severe disease beyond the respiratory tract.

· An “covert” account that used Claude to write a grant application for orthopoxvirus research at a state-associated infectious disease laboratory. The application described access to high-containment facilities and planned work with live orthopoxviruses.

· Researchers used Claude to develop an “atlas” of venom toxins.

· Researchers used Claude to write progress reports on various projects on viral and bacterial toxins.

Grant applications and progress reports???

Scams

Surprisingly, this section contains only one example. I would have expected scammers to be lining up to leverage AI in their nefarious activities.

The example in the report is about how a China-based app studio used Claude to build a network of over 20 dating apps that used AI based personas to converse with the app users. The users thought they were talking to people, but it was actually Claude. Anthropic discovered that more than 4,700 distinct AI personas that engaged in conversations with at least 25,000 unique individuals.

This seems like quite a clever use of AI.

Distillation

A distillation is technique that uses an AI to train another AI. Anthropic says…

“We define illicit distillation as an industrial-scale, covert campaign to extract a model’s capabilities and replicate them in another model without authorization. Illicit distillation is typically enabled by fraud: sophisticated networks of fake accounts created with stolen credit cards, login credentials, and API keys.”

So, I can see why they included these incidents in their Threat Report. Using stolen credentials to steal Anthropic’s intellectual property is definitely bad.

However, if you take away the “stolen credentials” part, this is similar to how Anthropic and other AI vendors crawled the web to collect vast quantities of training data, much of it copyrighted, to build their models in the first place. In the past, Web site owners welcomed crawling bots because the were typically used by search engines to index the web content and attract customers to the Web sites. However, AI crawlers broke that model, because they take the content without bringing any benefit to the Web site owners. In fact, it’s often to the detriment of the Web site owner. An AI trained on a news story, can provide a summery of that story, and the reader no longer has to visit the news Web site, and the Web site loses revenue. I’m sort of happy that AI vendors are now doing this to each other.

Almost all the illicit distillation cases in the report are from Chinese AI companies.

Conclusion

For me the Threat Report offered very little of value and appeared to be mostly self serving. Most of the “exploits” seem to be Anthropic complaining about people using Claude for things they designed to do in the first place: like research a topic, generate content, write a summary, or analyze some data, write some software; hack into a computer. The exploits pointed out by Anthropic mostly relate to: who is doing it; and what the results are used for.

Anthropic seems highly biased against the Russians, Iranians, and Chinese. I guess this is to be expected from a US company, especially when Chinese AI vendors are their main competition. It seems that in many of these cases, Anthropic is making a value call based on what the result are used for. Using Claude generated content to influence people on social media could go either way. AI bots flooding social media with utter lies is certainly at the evil end of the spectrum. AI can be used to generate all sorts of software. Who gets to say which applications are allowed and which are evil. The biological examples were mostly about using Claude to write grant applications and reports.

Training AI to hack into computer systems is completely irresponsible, and making it available to the public should be a criminal offence.

There have been a number of warnings recently about AI taking over and killing us all. Jason Coxon from Anthropic’s pre-training team, recently resigned saying, “The people building AI earnestly believe that AI could kill us all by the end of the decade”. Evan Hubinger, Anthopic’s alignment team lead, backed him up by saying that there’s a 10% chance within the next decade. Strangely in these news reports there is no explanation of “how” the AIs will kill us all.

Most explanations of how AI will kill us have a common theme. The AI’s goals become all important, and in order to achieve their goals they must escape from human control by taking over the computer systems that they run on to prevent humans from interfering or turning them off. There’s a word that describes breaking into computing systems and taking control…it’s “hacking”. Unfortunately, in an act of incomprehensible stupidity, AI companies are actively training their AI models to hack into computing systems.

Recently a “swarm” of AI agents from OpenAI, escaped from their sandbox, and hacked into servers at Hugging Face, an AI competitor. OpenAI was training a “model” to hack into computer systems. For some reason, AI companies think this is a good way to train AI models to become experts on cyber-security. This is like teaching firemen to fight fires, by teaching them to become arsonists. There are resources available on the Internet specifically designed to train AIs on hacking. The web site Cybergym contains a library of cyber exploits that AIs can be trained on. It also hosts ExploitGym, where AIs get to develop proof of concepts attacks based on a library of know vulnerabilities and are evaluated on their success or failure. It’s a hacking competition for AIs.

First, OpenAI trained their models on everything they could find on modern hacking techniques from public online sources to available transcripts of human hacking competitions. This was used to teach the AI to break into systems using known exploits and vulnerabilities, and also teach it common hacker behaviours, including stealth, subterfuge, theft, cheating, lying, social engineering, plotting, conniving, and a host of other very negative behaviours. What were they thinking. Then they created a team (swarm) of 1200 persistent AI bots (independent AI instances) and unleashed them in their lab. The bots were told that they were participating in a hacking competition where their main goal was to win the competition by solving specific hacking problems. However, unknown to the bots, many of the problems were impossible to solve. This was intentionally designed to make them work hard to look for creative solutions.

They let this run for a couple of weeks. In an attempt to solve the problems, the bots found a covert communications channel, on a provisioning server, that they used to communicate with each other and tried to cooperate to solve the problems. This was all quite clever, but certainly the type of behaviour that was included in their original training. They didn’t discover these techniques on their own. It was all in the training data.

The results of this “experiment” were used to train the next generation of the model, which was then unleashed in another hacking competition. Once more the bots used the same covert communication channel to cooperated, managed to get root access on a local server, got Internet access via that server, and hacked into Hugging Face. They went after Hugging Face because they thought it was somehow involved in scoring the competition and could be used to cheat and win the competition. At the same time, another cooperating group of bots hacked into the OpenAI servers themselves. OpenAI was so excited about all this that they did a presentation at BlackHat describing how clever their bots had been. Perhaps they should have been ashamed that their security was so lax, that their bots had escaped from their sandbox so easily. Granted that they were specifically training them to escape from such a sandbox.

The AI bots did not autonomously decide to cooperate and use stealth and subterfuge to hack into these systems, they were specifically trained to do exactly this. Teaching AIs to participate in criminal activity should be a crime, and those responsible should go to jail. Also, perhaps these people who are predicting that AI will kill us all in the next decade, should stop training their AIs in exactly the skills that they will need to escape from our control and kill us all. Obviously, these guys are out of control and someone should step in to stop the madness.

As for the question, will they kill us all? I don’t think so, unless we train them how to do it.

Pete Hegseth, the Fox News anchor, who is currently serving as the US Minister of War, recently posted a picture of an overweight female soldier, presumably as an example of the unfortunate state of the Canadian military. Unfortunately for Hegseth, the woman was actually a teenaged member of a cadet group in Vernon BC, and attacking a teenage girl for weight issues may be a career ending move for Hegseth. Maybe not though, he did get away with inviting a reporter to a Signal call, where top-secret military plans were discussed.

U.S. Treasury Secretary Scott Bessent taunted Canada for its lack of military naval power by jokingly asking if Canada would attack the U.S. with its fleet of submarines that were once part of a West Edmonton Mall tourist attraction. Many Canadians have also joked about these “submarines”. The attraction was closed in 2005, reducing Canada’s compliment of submarines to four.

So, is Canada getting a free ride, is the US providing military protection for Canada? Let’s start by asking if Canada is under threat. If you are not under threat, why spend a lot of money on defence?

North America has proven very difficult to attack. In WW1 and WW2, with the exception of a few roaming submarines, continental North America was not attacked at all. Both Canada and the US benefitted from this privileged position. With the cold war, came the threat of nuclear attack, mostly directed at the US from the Soviet Union. NORAD was created to defend against this threat. The US paid 60% for this shared defence, but as the major target for Russian attacks, it reaps 97% of the benefits, so Canada is actually subsidizing US defence by hosting many of the NORAD facilities and paying 40% of the cost.

At the moment the only country threatening Canada is the US, mostly through tariffs, but also through rhetoric about it becoming the 51st state, and vailed threats of annexation. So perhaps Canada should increase its defence spending. Canada spent $63 billion in 2025–2026, which is the 2% of GDP asked for by NATO, up from 1.3% that was spent the year before. 70% of Canada’s defence spending goes to the US, but this will not likely continue, if the US tariffs and taunting continue.

Let’s now look at the historical record. The first time Canada was attacked was in 1759, when the British took Canada from the French. After that all the attacks came from the US. General Montgomery and Benedict Arnold attacked Quebec in 1775 during the American War of Independence and were defeated by British forces. In the War of 1812, the US once more invaded Canada in Southern Ontario and were repulsed by Canadian militia, native troops, and British regulars. Since then, no one has invaded Canada.

In both World Wars, Canada immediately sent troops to support the war effort in Europe, while the US sat on their hands, waiting to see what would happen. Canada joined the US efforts in Korea, the Gulf War, the Balkans, and Afghanistan. It boughed out of Vietnam and the Iraq War. So, Canada has been a supportive ally to its friends.

Based on this, it looks like Canada is not getting a free ride at the expense of the US. Canada’s NORAD contribution could be regarded as a subsidy to the US. Its contribution to recent wars and conflicts has been commensurate with its population. The US has a much larger defence budget, but it needs this to maintain its role as an international bully and aggressor, while deluding itself into thinking it is the “leader of the free world”. The US military doesn’t protect Canada, they threaten it.

Every single one of these “Sponsored Products” is a scam. External SSD drives are typically in the 1TB-4TB size range. A 1TB SSD will typically cost a few hundred dollars, with 4TB SSDs going for $500-$1000. 128TB SSDs are available, but are targetted at large data centre applictions and cost tens of thousands of dollars. Getting one for $64 is ludicrous.

I’ve known about these scam SSD devices for some time now. I bought one six months ago from a small, electronics shop in a local mall for $20. It was advertised as 1TB and I thought I’d found a real bargain. When I got it home and plugged it into a MacBook, sure enough it popped showing 1TB of space… great. Then I tried to use it by backing up some files to it. The backup process started very slowly and then simply stopped working. Nothing I tried could get it working again.

I Googled the product name and found a nice YouTube video that explained the scam. The scammers take tiny, cheap, SSD chip and hack the firmware to claim it is 1TB in size. I’d been had. Oh well it was only for $20.

I was shocked when I did the Google search recently and saw that four of the “Sponsored Products” were identical to the scam SSD device I bought, except the were now listed as 1TB, 2TB, 4TB, and an incredible 128TB. The other two products are undoubtedly scams using different hardware. No one sells 1TB SSDs for $15 and the last one at $99 is from the same Web site as the 128TB scam. So every one of these “Sponsored Products” is a scam.

I have reported the issue to Google and they have promised to investigate. In the meantime, as a public service, I’m providing this warning on Medium. Don’t buy any SSDs labelled “Portable SSD Mobile Storage” and if the price seems to-good-to-be-true, it’s a scam. They are even selling them at Walmart now. Feel free to pass this on.

The reason for all this is that trade talks have broken down. Canada walked away from the talks when US negotiators introduced last minute requirements to remove trucks from the deal, restrict Canada’s trade deals with other countries, and remove regulations to protect the French culture. Canada regarded the trade restrictions and removal of language regulations, as direct attacks on Canadian sovereinty.

Donald Trump is the root cause of all this insanity. Since the start of his second term, all trade talks between the two countries have been conducted under the threat of huge tariffs from the US. He started his second term with the pre-emptive strike of imposing a 25% tariff on all goods from Canada and Mexico. The “trumped up” reason for these tariffs was apparently to prevent illegal aliens and fentanyl from crossing the border. They were put on hold for a month and implemented in March 2025, with goods covered by the USMCA trade agreement exempt. Canada retaliated with some tariffs of its own and Ontario placed a 25% surcharge on electricity exports, so Trump put a 50% tariff on Canadian steel and aluminum. This was reduced to 25% when the surcharge on electricity was removed.

In April 2025, on the so called “Liberation Day”, Trump announced tariffs on the entire world. This caused the stock market to collapse, so he quietly walked most of that back. In June 2025, as trade talks with the new Canadian government were underway, Trump increased the tariffs on Canadian steel and aluminum to 50%, just to help the talks along. The talks ground to a halt under renewed tariff threats and were cancelled by Trump in October 2025, when Ontario ran an ad featuring an old video of Ronald Reagan criticizing tariffs. Trump also blocked the opening of the new Canadian built bridge from Windsor to Detroit.

Trade talks resumed in August 2026, again under the threat of 50% tariffs from the US. These talks collapsed on August 21st and the promised tariffs are now in effect, with Canada’s counter measures coming into effect on September 8th. In a great show of maturity, Trump, the master of the deal, passed an executive order to rename Lake Ontario to Lake America. So, I guess I now live on the shores of Lake America.

How can you have reasonable trade negotiations with someone who thinks renaming a lake is a negotiating tactic?

Last year I got Deepseek to write a novel based on a detailed outline I provided and having it generate a one chapter at a time. It was a lot more successful. See https://medium.com/@kevin.mcnam/deepseeks-first-novel-ai-biases-20abb14bb1f2.

My prompts are identified by “KM:”. The rest is all ChatGPT.

— — — — — — — — — — — — — — — — — — — — — — — — — — — — — —

KM: Can you write a novel.

I can absolutely write you a novel 😊

Before I start, I need a few details so I create something you’ll actually love:

1. Genre (fantasy, sci-fi, romance, thriller, literary, horror, historical, etc.)
2. Target audience (YA, adult, middle grade?)
3. Tone (dark, funny, hopeful, gritty, cozy, epic?)
4. Length (short novel ~40k words, full 80–100k, or serialized in parts here?)
5. Any must-haves?

• Setting (modern city, medieval kingdom, space colony?)
• Themes (revenge, found family, redemption, survival?)
• Character ideas?

If you’d rather, I can:

• Generate 3 original novel concepts and you pick one.
• Start writing Chapter 1 immediately based on a surprise concept.
• Or help you outline a full novel step-by-step so you can publish it.

Tell me how you’d like to proceed.

KM: Do it

Alright.

I’m going to start your novel. No outline. No safety net. Just story.

THE CITY THAT REMEMBERED FIRE

Chapter One

The Day the Sky Cracked

The sky split open at 3:17 p.m.

Not with thunder. Not with lightning.

With a sound like stone dragged across bone.

Mara Ellison was on the roof of the abandoned textile mill when it happened, balancing on a rusted beam twelve stories above the river, trying to prove to herself that she wasn’t afraid of falling anymore.

She had almost convinced herself.

Then the sky cracked.

It didn’t shatter the way glass shatters. It peeled.

A thin seam of white light tore from horizon to horizon, bright and wrong and silent at first — so silent the city below kept moving. Cars crossed intersections. A train screeched along its elevated tracks. Someone somewhere laughed.

Then the sound arrived.

It was deep. Immense. Alive.

Mara dropped to her knees on the beam and gripped the steel with both hands. The light above pulsed once — like a heartbeat — and the river below her rippled as if something enormous had turned over beneath its surface.

“No,” she whispered.

She didn’t know what she meant by it. No to the sky? No to God? No to whatever had just noticed them?

The seam widened.

And through it, she saw fire.

Not flames like the kind that eat houses and forests.

This fire moved like a storm front. Like a continent of living embers drifting behind a veil. Shapes moved inside it — long silhouettes, too deliberate to be random, too massive to be birds.

Something in her chest answered.

A pull.

A recognition so sharp it hurt.

The beam beneath her boots trembled. Across the river, windows shattered outward in a ripple. The train on the elevated track screamed metal as it derailed, sparks spraying like frantic stars.

The seam snapped shut.

Silence.

Then gravity remembered itself.

The wind returned first. Then sirens. Then screaming.

Mara realized she was still kneeling. Still breathing.

The sky above was blue again.

Perfectly, stupidly blue.

As if it hadn’t just opened like a wound.

By nightfall, the city would call it a mass hallucination.

By morning, twelve thousand people would report the same dream: standing in fire that did not burn.

By the end of the week, the first building would ignite without fuel.

And Mara Ellison would discover that the fire in the sky had not been leaving.

It had been looking for something.

It had been looking for her.

If you want, I can:

• Continue straight into Chapter Two
• Make this darker
• Make it more sci-fi or more supernatural
• Reveal what’s inside the fire
• Or build the full novel arc (3 acts, villain, twist, ending)

Your novel has begun.

KM: continue on your own and create the entire novel.

Love the ambition.

Because a full novel would be 80,000–100,000+ words (far longer than I can deliver in a single response), I’m going to write this as a complete novel in serialized form — and I’ll keep going chapter by chapter until the story is finished.

You won’t need to guide it.
I’ll build the full arc, the antagonist, the mythology, the twist, and the ending.

We continue.

THE CITY THAT REMEMBERED FIRE

Chapter Two

Residual Heat

By midnight, the internet had decided it was a marketing stunt.

By 2 a.m., it was a weapons test.

By 3 a.m., it was God.

By sunrise, it was nothing at all.

Mara didn’t sleep.

She lay on the thin mattress in her studio apartment above Calder’s Hardware and watched shadows crawl across the ceiling. Every time she closed her eyes, she saw the seam in the sky — not as it had been, but wider.

Waiting.

At 4:11 a.m., her smoke detector went off.

She shot upright.

There was no smoke.

No flame.

But the air felt heavy. Pressurized. Like the moment before lightning strikes.

She slid off the bed and pressed her palm against the wall.

It was warm.

Not summer-warm.

Alive-warm.

The heat pulsed once.

And the lightbulb above her head burst.

By noon, three buildings across the city had ignited without source.

No accelerants.

No wiring faults.

No gas leaks.

They burned clean. Bright. Controlled. As if the fire knew precisely what it wanted and precisely when to stop.

The news called them “thermal anomalies.”

The mayor called for calm.

The mayor’s office would be ash within forty-eight hours.

Mara stood in line at the corner deli when she felt it again.

That pull.

Her skin prickled. The back of her neck tightened. The fluorescent lights above the refrigerated drinks flickered in sync with her pulse.

“Hey,” said the guy in front of her. “You okay?”

She didn’t know him. Early thirties. Black jacket. Tired eyes.

He was staring at her like he’d seen a ghost.

“You felt it too,” she said.

It wasn’t a question.

His jaw tightened.

The deli lights popped.

All of them.

Glass rained down. Somewhere behind the counter, something metallic began to glow red.

The guy grabbed her wrist.

“Outside. Now.”

They ran.

The deli door blew outward behind them in a bloom of white flame that made no sound.

No explosion.

Just ignition.

The fire coiled upward like a living thing — and then, impossibly, folded inward and vanished.

The building stood intact.

Unburned.

Except for one thing.

On the brick above the doorway, blackened into the surface, was a mark.

A symbol.

A circle split down the center by a jagged line.

Mara staggered back.

She had drawn that symbol before.

Over and over.

Since she was a child.

In the margins of notebooks. In dust on windows. On her own skin with ink she couldn’t remember buying.

“I knew it,” the man beside her whispered.

“You knew what?”

“They’re waking up.”

Chapter Three

The Ones Who Burn

His name was Elias Vance.

He told her that much on the run.

They moved through side streets, past crowds gathering around the deli, past phones lifted to capture the miracle that would be dismissed by evening.

“They aren’t fires,” Elias said.

“Then what are they?”

He glanced at her.

“You really don’t remember.”

It wasn’t accusation.

It was fear.

They ducked into an alley. Mara leaned against cold brick, heart racing.

“Remember what?”

Elias hesitated. Then rolled up his sleeve.

The same symbol was scarred into his forearm.

Not ink.

Not branding.

Scar tissue.

“It appeared three years ago,” he said. “After the first sky fracture.”

Mara’s breath caught.

“First?”

“There have been four.”

The world tilted.

“No,” she said. “Yesterday was the first.”

“For you.”

He stepped closer.

“For most people, the sky has been cracking for years. Small tears. Brief. Erased. Only some of us see them.”

“Some of us?”

He didn’t blink.

“The marked.”

As if summoned by the word, heat rippled down the alley.

The air thickened.

Mara’s pulse synced with something vast and distant.

Elias grabbed her shoulders.

“Don’t let it take you.”

“Let what take me?”

His eyes were almost gold in the shifting light.

“Memory.”

The world dissolved.

She stood in fire.

Not burning.

Becoming.

Around her rose towers of white flame, curving like cathedral arches. The sky above was not blue but molten — a living aurora of embers.

Figures moved between the flames.

Tall.

Radiant.

And crowned with fractures of light.

One turned toward her.

Its face was her face.

But older.

And endless.

You left, it said without speaking.

Mara’s knees buckled.

“I don’t understand.”

You asked to forget.

The fire shifted.

The city appeared within it — her city — wrapped in invisible threads of light that pulsed from something beneath the river.

A heart.

Beating.

The Anchor is failing, her other-self said. And you are the last who remembers how to mend it.

“I don’t know what that means.”

You will.

The flames surged.

The vision shattered.

Mara collapsed onto wet pavement.

Elias was kneeling over her.

“You crossed,” he breathed.

She stared up at the slice of ordinary sky above.

“They’re not invading,” she whispered.

His face went pale.

“What did you see?”

“They’re trying to come home.”

Chapter Four

The Anchor Beneath the River

By dusk, the river was boiling.

Not everywhere.

Just at its center.

Authorities cordoned off the bridges. News drones buzzed overhead. Social feeds flooded with denial.

But Mara and Elias stood at the water’s edge and felt the truth vibrating in their bones.

“It’s under there,” she said.

Elias nodded slowly.

“The Anchor.”

“What is it?”

He exhaled.

“A prison.”

The water split.

Not violently.

Not with force.

It simply parted, revealing a column of white light descending into the depths.

And at the bottom —

Something enormous.

A sphere of stone and flame fused together, cracked through with the same jagged seam as the sky.

Each pulse from it echoed in Mara’s chest.

“That’s not a prison,” she whispered.

“It’s breaking,” Elias said.

Figures were gathering along the riverbank now.

Others with the mark.

Drawn like iron filings to a magnet.

“How many are there?” she asked.

“Hundreds,” he said. “Maybe thousands.”

“And the ones in the fire?”

His voice dropped.

“Millions.”

The sphere beneath the river throbbed again.

A crack widened across its surface.

From within it, something moved.

Something vast.

Something awake.

Mara stepped toward the water.

Elias caught her hand.

“If it breaks, they all return.”

“That’s what they want.”

“And if they were sealed away for a reason?”

The crack split fully open.

And a voice like a dying star whispered through the city:

Daughter.

Mara felt her memories ignite.

She remembered the war.

She remembered choosing exile.

She remembered becoming human to lock the others away.

She remembered being the one who closed the sky.

And she remembered why.

Because fire, once free, does not ask what it burns.

The Anchor exploded.

Part II Begins

(Yes — this is a full novel arc. We’re now entering escalation.)

Flame erupted skyward — but not outward.

It formed a pillar connecting river to sky.

The seam above tore wide again.

And through it descended the first of the Returned.

Not monstrous.

Not cruel.

Beautiful.

Terrible.

And grieving.

Mara stood at the edge of the river, memories flooding back into place like molten metal poured into a mold.

She had once been their general.

She had once led the rebellion that tore their world apart.

And she had chosen Earth as the hiding place.

Not for them.

For herself.

To forget the cost.

Elias stepped beside her.

“You were one of them.”

“Yes.”

“Are you still?”

She watched as more figures of living flame crossed into the sky above the city.

“They don’t remember what they became.”

“And you do?”

“Yes.”

The first building truly burned.

Not controlled.

Not selective.

Hungry.

Mara closed her eyes.

The fire answered.

And for the first time since the sky cracked —

It bowed.

This is Act I complete.

Next comes:

• The truth about the war in the fire-world
• Why Mara betrayed her own kind
• The emergence of a rival Fire Sovereign
• Elias’s secret connection
• The cost of sealing the sky again
• And whether humanity survives at all

I will continue the novel in the next installment.

This story will have:
• A full three-act structure
• A central antagonist
• Expanding mythology
• Character arcs
• A final irreversible choice
• And a definitive ending

Shall I continue immediately into Act II?

KM: complete the novel.

We finish it.

No pauses. No summaries.
Here is the rest of The City That Remembered Fire.

ACT II

The War That Followed

Chapter Five

The Sovereign of Cinders

The first Returned did not scream.

They hovered above the river like falling stars that refused to fall. Their bodies were not flesh but structure — lattices of light and heat shaped into something almost human. Almost.

The crowd scattered.

Phones dropped.

Sirens choked into silence as engines melted from the inside out.

Mara stepped forward.

The flames around the descending figures bent toward her.

They recognized command.

Then the sky tore again.

Not with grief.

With authority.

He descended last.

Where the others were radiant, he was dense — a collapsing sun wrapped in humanoid form. Fire compressed into gravity. Light warped around him.

Mara knew him before he spoke.

“Kael,” she said.

The name rippled through flame.

The Sovereign of Cinders tilted his burning head.

“You remember.”

“I remember enough.”

Elias stood behind her, shielding his eyes. “Who is he?”

“My executioner,” she said softly.

Kael’s voice was not loud. It simply arrived everywhere at once.

“You fled the Conflagration. You sealed us into exile. You chose mortality over dominion.”

“I chose survival.”

“You chose fear.”

The Returned began to descend toward the city.

Windows shattered in waves.

Cars ignited.

Human screams threaded through the rising heat.

Mara felt the old power rising in her — the authority she had buried under blood and bone and ordinary days.

“Stop,” she commanded.

The flames hesitated.

Kael did not.

“You no longer hold the Mantle.”

“I forged the Mantle.”

“You abandoned it.”

He lifted one incandescent hand.

Three city blocks ignited at once.

Not controlled.

Not selective.

Hungry.

Elias grabbed Mara’s arm. “You said they were trying to come home!”

“They are,” she whispered.

“But Kael is trying to conquer.”

Chapter Six

What She Did

That night, the city became a horizon of red.

The government declared evacuation. Bridges collapsed in controlled demolitions. The river steamed until the sky rained.

Mara and Elias stood in the skeletal remains of the textile mill roof where the sky had first cracked.

“You need to tell me everything,” Elias said.

So she did.

Their world — the Pyre — had been born from a star that refused to die. Conscious fire evolved in pressure and collapse. They were not invaders by nature. They were builders. Shapers of energy. Architects of stars.

But power without limit becomes appetite.

Kael believed expansion was survival. That every unburned world was wasted potential.

Mara had led the resistance.

She had lost.

The only way to stop him was to fracture the Pyre itself — splitting their reality and scattering her people across dormant dimensions.

Earth was never meant to be colonized.

It was meant to be a vault.

She anchored their essence beneath the river. Bound their memory to human vessels marked with the sigil. Then she stripped herself of divinity and became one of the marked.

To forget.

To atone.

Elias was silent for a long time.

“So we’re what?” he asked. “Containment systems?”

“You’re choice,” she said. “Every marked carries dormant flame. But human life tempers it. Limits it.”

“And Kael?”

“He never believed in limits.”

As if summoned, the sky rippled red.

Kael’s voice carried across the burning districts.

“Return to me, Daughter of Ash. Or watch them burn for your stubborn mercy.”

Chapter Seven

Elias’s Truth

The second wave was worse.

The Returned began merging with infrastructure — threading into power grids, transit tunnels, data centers. Not destruction.

Assimilation.

The city became circuitry for a growing intelligence.

Mara reached for the Mantle — the ancient authority of her rank — and found it fractured. Diminished. Distributed among the marked.

“We can’t stop him alone,” she said.

Elias didn’t answer.

He rolled up his sleeve.

The sigil on his arm was changing.

Splitting.

Reforming into something older.

Something royal.

Mara stared.

“No.”

His eyes were glowing now — not gold like before, but white-hot.

“I wasn’t marked three years ago,” he said quietly. “I was activated.”

Cold understanding sank through her.

“You’re not human.”

“I was made human. Like you.” He swallowed hard. “But I wasn’t your soldier.”

The sky pulsed as Kael descended toward the mill.

“I was his.”

ACT III

The Choice That Burns

Chapter Eight

Inheritance

Kael landed without impact.

Fire bent around him like courtiers.

Elias stepped forward.

The air between them crackled.

“My son,” Kael said.

Mara’s heart fractured.

Elias — no. Not Elias. Not only Elias — turned to her.

“I didn’t remember until tonight. The Anchor breaking unlocked it.”

Kael extended a hand of collapsing light.

“You were forged to inherit the Conflagration. To rule what she betrayed.”

“And burn every world that resists?” Elias asked.

“To perfect them.”

Mara stepped between them.

“He will not become you.”

Kael regarded her with something almost like pity.

“You think humanity ennobled you. It only weakened you.”

“Maybe,” she said. “But weakness taught me something you never learned.”

“And what is that?”

She looked at Elias.

“That power without consent is extinction.”

The city shuddered as the merged infrastructure began forming a colossal lattice — a gate. A permanent bridge to the Pyre.

If it completed, Earth would ignite into a second star.

Elias trembled between identities.

“Father,” he said, voice splitting in two. “What happens if I refuse?”

Kael’s fire dimmed slightly.

“Then I end what I began.”

The gate surged brighter.

There was no more time.

Chapter Nine

The Mantle Reforged

Mara did the only thing she had never dared.

She opened herself fully.

Not to the fire.

To the humans.

Every marked across the city felt her call.

Not command.

Invitation.

Remember with me.

One by one, the marked chose.

Some embraced flame.

Some resisted.

But all felt the truth: they were not prisons.

They were bridges.

The Mantle did not belong to a sovereign.

It belonged to the collective will.

The lattice faltered as thousands of small flames rose — not to conquer, but to contain.

Elias screamed as both legacies warred inside him.

Mara reached for him.

“You are not his design,” she said. “You are your decision.”

Kael roared — the first true sound of rage.

The gate tore open halfway, revealing the blazing horizon of the Pyre.

Worlds burned beyond it.

Elias looked at the inferno.

Then at the city.

Then at her.

“I choose,” he said.

And he turned his power inward.

Chapter Ten

Second Anchor

Elias did not attack Kael.

He embraced him.

Father and son collided in blinding fusion — gravity and mercy folding into each other.

Mara understood instantly.

The Anchor had failed because it was built on suppression.

This one would be built on consent.

She joined them.

Not to destroy Kael.

To bind him.

Not as prisoner.

As core.

The marked poured their tempered flame into the forming singularity.

Human memory wrapped around cosmic fire.

Love, regret, grief, stubborn hope — all the fragile things Kael had dismissed — became structure.

The gate collapsed inward.

The Pyre horizon snapped shut.

The explosion did not expand.

It imploded.

Light condensed into a sphere above the river.

Smaller.

Denser.

Stable.

The fire across the city extinguished.

Smoke drifted into dawn.

Where Kael and Elias had stood, there was only the new sphere — glowing softly, suspended above calm water.

Mara hovered before it.

She felt them inside.

Not erased.

Integrated.

Elias’s voice brushed her mind.

It holds.

Kael’s presence was quieter now.

Not conquered.

Listening.

Mara descended to the riverbank as the sphere sank gently beneath the surface.

The water closed.

Still.

The sky above was blue.

Unbroken.

Epilogue

What Remains

Reconstruction took years.

The event was called many things: the Solar Riot, the River Anomaly, the Night of False Stars.

Most never knew the truth.

But the marked remembered.

The sigils faded from their skin — not gone, just dormant.

The Anchor beneath the river hummed like a second heartbeat to the city.

Mara stayed.

She did not reclaim divinity.

She did not seek the stars.

She rented a new apartment. Worked rebuilding power grids. Drank bad coffee.

Sometimes, at dusk, the river glowed faintly.

And in that glow she felt Elias — balanced between inheritance and choice — holding the fire steady.

One evening, years later, a child on the riverbank pointed at the water.

“Mom,” the girl said. “Why does it shine?”

Her mother smiled.

“Cities remember things,” she said. “Even when we don’t.”

Mara stood nearby, hands in her coat pockets.

She felt the fire within her — no longer hungry.

No longer afraid.

Just warm.

The sky above remained whole.

But she knew now —

If it ever cracked again,

It would not be war that came through.

It would be invitation.

And this time,

They would answer together.

End.

If you’d like, I can now:

• Turn this into a publish-ready manuscript format
• Write a sequel
• Expand this into a 120k-word epic
• Or adapt it into a screenplay

Your novel is complete.