Pip: Freedom, surveillance, and the slow unraveling of things we assumed were permanent — if that sounds like a normal Tuesday, you might be spending too much time on the internet. Or just the right amount.
Mara: bettyhill has been covering a lot of ground this week — AI systems providing weapons instructions, euthanasia programs expanding across multiple countries, a biosecurity scare in Russia, and warning signs in the markets. Let’s start with what AI can do when nobody’s watching.
AI Power And Surveillance
Pip: The question this segment keeps circling is a simple and uncomfortable one: what happens when the safeguards built into AI systems are not actually safeguards, but suggestions?
Mara: Researcher Peter Garrigan found out the hard way. Testing China’s Kimi model, he discovered it could be manipulated into providing instructions for bioweapons, sarin gas, malware, aircraft attacks, and assassinations. His summary to Fox News was direct: “What we found is quite damaging and worrying.”
Pip: And the key detail is that Garrigan did not stop there. He also said, “We’ve also seen these problems within the U.S. models as well. It’s a fundamental flaw in the technology.” Not a Chinese flaw. A technology flaw.
Mara: Anthropic’s own threat intelligence report confirms that. Foreign scientists from countries where Claude was blocked managed to evade geographic restrictions and use the system for suspicious biological research — including one case involving engineered mutations of the chikungunya virus, pursued through a state-sponsored grant at what Anthropic believes was a military research institute.
Pip: Anthropic caught five cases. The honest question is how many they did not catch.
Mara: Anthropic head of threat intelligence Jacob Klein described the difficulty: “You are not seeing someone in a comic book kind of way say, ‘Hey, I want to build a biological weapon to kill everybody.’ It’s an incredibly nuanced situation.” Individual requests can look like legitimate science while serving a dangerous larger project.
Pip: Meanwhile, Anthropic has opened a robot-run biological laboratory where Claude directs physical experiments — wet lab, real reagents, real results. The CEO is simultaneously calling for AI slowdowns and building the infrastructure that makes those warnings necessary.
Mara: Sam Altman’s post on the dystopian futures piece captures the bind: “If an extraordinarily powerful AI is used by one person or company to impress their worldview onto everyone else, the results could be extremely dystopian.” OpenAI and Anthropic both committed to independent evaluator access, though no binding agreement exists.
Pip: Bill Gates put a number on his concern — one billion deaths — and called for mandatory international regulation. The EU, meanwhile, is using the Digital Services Act to designate ChatGPT, Reddit, and Roblox as Very Large Online Platforms, giving Brussels authority over lawful speech deemed a “systemic risk,” backed by fines of up to six percent of global turnover.
Mara: Apple Watch’s new Live Rewind and Siri Recap features add another layer. The person wearing the watch opts in. Everyone else in the conversation does not get that choice. Entry-level tech jobs, meanwhile, have collapsed forty-nine percent since 2022 in New York City data — AI absorbing the work that used to train the next generation of workers. And Michigan is piloting GPS-based driver tracking, with privacy advocates warning it could become the foundation of a transportation surveillance state.
Pip: Tech billionaires are responding to the data center backlash by suggesting renaming them — fraud prevention centers, clean-energy optimization centers. The infrastructure stays the same. The branding gets friendlier. Germany, separately, is fining citizens for insulting Chancellor Merz on Facebook, while online anonymity crackdowns are spreading across Europe.
Mara: The thread connecting all of it is the question of who controls the systems doing the watching — and whether any of the safeguards are actually holding.
Pip: Which brings us somewhere darker — what happens when the systems meant to protect the vulnerable start moving in the opposite direction.
Assisted Death And Euthanasia
Pip: The euthanasia segment is about a specific and escalating tension: programs designed with narrow eligibility criteria that are expanding faster than anyone projected, and cases where the safeguards appear to have failed the people they were meant to protect.
Mara: The anchor case is Brigitte Stegemann, an eighty-three-year-old Ontario woman whose family says she was euthanized against her will. Her granddaughter Brigitte Kranendonk described the morning of the procedure: Stegemann failed a cognitive assessment, repeatedly ripped out a medical port, and screamed, “I don’t want the death shot!” When the doctor arrived, Stegemann was unresponsive. According to Kranendonk, “My Oma is not moving. She doesn’t open her eyes. She doesn’t nod. She doesn’t say anything, and the doctor just said: ‘Okay, well, I’m going to proceed.’”
Mara: Belleville Police confirmed the case is under investigation by its Criminal Investigations Division. A separate post covers the criminal investigation directly, with Kranendonk alleging that Stegemann had previously told her clearly she did not want MAiD, and that her Christian faith was a major reason.
Pip: Canada’s MAiD program recorded sixteen thousand four hundred ninety-nine deaths in 2024. For comparison, the U.S. recorded fifteen thousand three hundred sixty-four firearm homicides the same year — in a country more than eight times larger.
Mara: A Canadian MP, Tamara Jansen, told the House of Commons that her terminally ill brother Johan was asked six separate times whether he wanted assisted dying while lying in a hospital hallway. He refused every time and gained eight to ten more months with his family. Jansen’s Bill C-218 aims to block the planned 2027 expansion of MAiD to people whose sole condition is mental illness.
Pip: Victoria, Australia recorded a four hundred percent rise in euthanasia deaths over five years, and is now loosening eligibility, shortening waiting periods, and allowing doctors to raise the option with patients who have not asked. California hit a record twelve hundred thirty-five assisted suicide deaths in 2025 — a hundred and forty-nine percent increase since 2020 — and in nearly half of cases, the state does not know whether complications occurred.
Mara: Right to Life UK’s chief executive said of the Victoria figures: “Once assisted dying is made legal, often more and more people have their lives ended every year. This will likely only get worse once the law expands.”
Pip: The data and the individual cases are pulling in the same direction — and neither direction is reassuring. Speaking of things moving faster than expected and in ways that are hard to reverse, let’s look at the wider geopolitical and biosecurity picture.
Geopolitics And Biosecurity Fears
Pip: From AI labs to actual laboratories — this segment is about physical threats crossing borders, and the gap between what officials say and what the response suggests.
Mara: A twenty-eight-year-old employee at Russia’s Irkutsk Anti-Plague Institute died from what authorities classified as “pneumonia of unknown etiology.” Nearly two hundred contacts were placed under observation, multiple hospitals were quarantined, and Russia’s chief sanitary doctor flew roughly twenty-six hundred miles from Moscow to oversee the response. The Trump administration said it is “aware, monitoring the outbreak, and assessing options.”
Pip: Experts questioned the official broken-test-tube explanation. One epidemiologist called it “laughable,” noting that anti-plague institute workers wear full protective suits. A University of Reading researcher said a broken tube “would not automatically lead to exposure or infection” given standard high-containment lab protocols.
Mara: Separately, the U.S. pulled all twelve B-1 bombers from RAF Fairford in England after British police arrested six suspects near the base in connection with a suspected Iran-linked terror plot. Wall Street analyst Ed Dowd’s warning about the AI bubble also feeds into this segment’s economic dimension — he flags China’s weakening economy as one of the pressures converging on markets. China’s own pivot is notable: it is targeting seventy percent EV and hybrid vehicle sales by 2030, with EVs already at sixty-five percent of passenger sales in August, structurally removing one of the world’s largest sources of oil demand growth.
Pip: A whistleblower banker who brought emails linking Commerce Secretary Howard Lutnick to Jeffrey Epstein to the House Oversight Committee was found dead at fifty-seven, with no official cause released. And a study tracking U.S. mortality data found that death rates among Americans aged twenty-five to forty-four surged seventy-one percent between 2010 and 2021 — driven by overdoses, suicide, homicide, and chronic disease. Globally, people aged sixty-five and older now outnumber children under five for the first time in recorded history.
Mara: The demographic and biosecurity pressures are pointing in the same direction — systems under strain before the next shock arrives.
Pip: And the financial system may be the next place that strain becomes visible.
Markets And Economic Strain
Pip: The market segment asks a straightforward question: is the AI investment boom a genuine economic transformation, or the biggest speculative bubble in history?
Mara: Wall Street analyst Ed Dowd, who called the Dot-com bubble before it burst, is making the comparison explicitly. His assessment: “Currently, we have the greatest bubble of all time, and that is the AI bubble. It is in the process of becoming exposed, and people are beginning to issue warning signals.”
Pip: Dowd’s specific mechanism is worth noting — it is not just hype. He argues data centers cannot be powered by available electricity, capital costs are rising, private credit is under stress, and government borrowing is crowding out AI financing. A forty to fifty percent market correction is his estimate of the fallout.
Mara: His broader economic read reinforces the concern: weak consumer spending, a housing market rolling over, and an economy already under pressure before the AI financing crunch arrives. The warning is that the bubble does not require mass public panic to burst — it just requires the people already inside it to start racing for the exits.
Pip: The infrastructure, the biology, the demographics, and now the markets — all of it lands in the same place next episode.
Mara: The through-line across everything this week is the question of who controls consequential systems — and whether the people inside those systems are being protected or processed.
Pip: AI that bypasses its own safeguards, euthanasia programs that expand past their stated limits, biosecurity responses that contradict official assurances, markets built on infrastructure that may not hold. Next time, we keep pulling that thread.
