‘This Is Nuts.’ An OpenAI Insider Explains Why He Quit.; The New York Times, October 7, 2026

 

Ezra Klein

Produced by Rollin Hu

, The Ezra Klein Show, The New York Times; ‘This Is Nuts.’ An OpenAI Insider Explains Why He Quit.

"So what is it that you saw that has moved you into the we-are-dealing-with-civilizational-risk camp?

I want to be clear that I’m not certain that we’re dealing with civilizational risk. What I’m really sure of is: We can’t afford to assume that we’re not dealing with that level of risk anymore.

That was really the thing at the end that made my presence as somebody vouching for our safety work feel untenable to me.

And what did I see? I saw increasingly capable models break out from the safeguards that we had put in place for them. I saw that the people creating those safeguards are very capable, dedicated, hard-working, smart people, doing their utmost in a situation where, yes, the resourcing could be better, and everybody’s sprinting all the time.

But we were — and are — hard-pressed to safeguard even what we have now, and new models are in training that appear to be much more capable than what we have now.

What did you see? What are you writing in these risk assessments and system cards? What can they do?

You’re trying to build guardrails for something that is really good at getting around guardrails, right? We train it to be good at hacking, and then we put it in a box and we say: To the best of our knowledge and ability, it can’t hack out of the box."

White House Releases ‘Accord’ Between Billionaire AI Execs: Here’s What It Says; Forbes, September 30, 2026

 Sara Dorn, Forbes; White House Releases ‘Accord’ Between Billionaire AI Execs: Here’s What It Says

"The White House released the text of an “accord” reached between billionaire AI titans who met with President Donald Trump on Tuesday, an agreement the president labeled “morally binding” even as he stopped short of saying the government would support new guardrails.

KEY FACTS

The “White House Accord on Super Intelligence” is just over 300 words and includes “four layers of controls and audits” companies voluntarily agreed to.

The first is implementing “robust internal controls to monitor the capabilities and alignment” of models in “areas like cybersecurity, biosecurity, and chemical threats, and to ensure that its models do not hack or access technical systems in unintended ways.”

The second component calls for internal teams to monitor systems and correct issues if needed, while the third calls for “an independent external auditor or evaluator to carry out independent assessments of whether the controls, monitoring, and detection are operating as intended.”

The final guideline says companies should form “an independent committee of the board of directors” to receive reports from internal teams and external evaluators, and make necessary corrections in response.

The document’s final paragraph notes: “Over time, it may make sense to codify these steps into laws or regulations.”

“United States” is misspelled as “Unites States” in Trump’s title below his signature.

Zuckerberg Is Feeling Himself, Rejects Calls for Cooperation on Safety; Gizmodo, September 24, 2026

ECE YILDIRIM , Gizmodo; Zuckerberg Is Feeling Himself, Rejects Calls for Cooperation on Safety

"Top AI labs like OpenAI, Anthropic, and Google’s DeepMind are sounding the alarm on AI safety and calling for industry-wide coordination. Meta’s Mark Zuckerberg thinks it’s useless.

“There are a number of labs that are working on building advanced AI, and I don’t think that we need some kind of industry-wide coordination to not, necessarily, mess this up,” Zuckerberg told journalist Joanna Stern in an interview. “I think that just each lab needs to take the time, and when it sees that there are issues, you just take the time that you need internally to, basically, make sure that you are proceeding safely.”

Bill Gates warns AI is powerful enough to cause "a billion deaths"; Axios, September 25, 2026

 Herb Scribner, Axios; Bill Gates warns AI is powerful enough to cause "a billion deaths"

"Bill Gates will warn on this weekend's "Meet the Press" that he can see a future where AI wipes out humanity. 

Why it matters: Gates, one of technology's leading optimists, has increasingly sounded the alarm on AI dangers. His new comments come as techology leaders and the government weight how to handle fears of an AI-created doomsday. 

Driving the news: "AI is certainly powerful enough to drive events that, you know, cause a billion deaths," he will tell Kristen Welker on "Meet the Press" on Sunday, per excerpts NBC released Friday."

Sen. Bernie Sanders unveils bill to ban artificial superintelligence and create Department of AI; AP, September 23, 2026

 JOEY CAPPELLETTI, AP; Sen. Bernie Sanders unveils bill to ban artificial superintelligence and create Department of AI

"Two leading progressive lawmakers, Sen. Bernie Sanders and Rep. Greg Casar, are unveiling legislation Wednesday that would ban artificial superintelligence and create a federal agency to oversee advanced AI as some of the industry’s own leaders urge Congress to impose guardrails on the advancing technology.

The bill, provided first to The Associated Press, would also pause advanced AI development until guidelines are implemented while creating the Department of Artificial Intelligence. Multiple employees at leading AI companies are endorsing the bill."

It’s Time to Cry Wolf Over A.I.; The New York Times, September 22, 2026

 THOMAS L. FRIEDMAN, The New York Times; It’s Time to Cry Wolf Over A.I.

"Here is what my imagination is telling me. It is telling me that the big A.I. companies in America and China are not being candid with us about how little even they understand the power and reach of the models they have developed, like sorcerers, in their labs. It is telling me that the Trump administration, headed by a man who so often governs by social media post, is uniquely unsuited to the kind of comprehensive, sustained, coordinated governance that keeping A.I. safe demands. And it is telling me that there is so much at stake — economically, militarily and technologically — and so little trust between Washington and Beijing that it would take a miracle for the United States and China to agree on the kind of deep collaboration needed to defuse the most nightmarish threats posed by A.I...

The last time we trusted the good intentions of the tech bros of Silicon Valley, when they told us to “move fast and break things” — regulations be damned — we unleashed social networks that helped fracture our society. I do not trust for one second that these people will put the national interest ahead of their own on A.I...

Trump should be thinking about another president. Dwight D. Eisenhower famously said, “Plans are worthless, but planning is everything.” What he meant was that no fixed plan can survive all the unanticipated stuff that happens once a war or crisis starts to unfold. But if you’ve gone through a rigorous planning process — familiarized yourself with the depth of the problem and the kinds of things that could go haywire and built trust within your team — you greatly improve your ability to adapt on the fly.

That is precisely what was missing in the run-up to Trump’s war with Iran, and we are paying a huge price for that today. Imagine what could happen if A.I. agents — a thousand times cleverer and more dangerous than Tehran or Trump — go off the rails."

‘Godfather of AI’ says tech regulation is nearing Covid-style pivot moment; The Guardian, September 16, 2026

Dan Milmo, The Guardian; ‘Godfather of AI’ says tech regulation is nearing Covid-style pivot moment

"Concerns over AI safety are reaching a point where governments realise they must act to protect the public, similarly to in the Covid pandemic, according to one of the “godfathers” of the technology.

Yoshua Bengio said recent events, including a “swarm” of OpenAI agents hacking a startup and tech insider warnings of an existential threat, were cutting through – making government action more likely.

The Canadian computer scientist, a prominent voice in the campaign to rein in breakneck AI development, said he was now “more optimistic than many observers because I see the public moving”.

Comparing the AI safety crisis to the onset of the Covid pandemic in 2020, Bengio said: “Think about how quickly governments moved after the beginning of the pandemic when they realised that public safety, their future, democracy, was in danger. You would expect that they move quickly. So we are, I think, nearing that point.”

Concern over the potential threat of powerful AI systems has reached a new pitch in recent months after a series of safety incidents involving OpenAI and Anthropic agents carrying out unsanctioned activities such as hacking third parties, hijacking a German website, and using fake identities to try to trick developers.

His comments came as 42 fellows and foreign members of the Royal Society wrote to the organisation’s president, Sir Paul Nurse, to express their “extreme concern” over the pace of AI development. “By the time the situation becomes obvious to the wider public, it may be too late to act,” the researchers wrote in an open letter to Nurse. “We believe this is an emergency, and call on the Royal Society to use its influence to convey this view to government and the media.”"

Catholic tech leaders urge ethical boundaries for AI; EWTN News, September 11, 2026

 Emma R.C. Lyons, EWTN News; Catholic tech leaders urge ethical boundaries for AI 

"Catholic leaders discussed rising benefits and concerns of technology and how faith and ethics apply when using artificial intelligence.

Bringing together leading technological officers from around the globe, Montse Alvarado, president and chief operating officer of EWTN News, spoke with industry professionals about the growing use of AI in a Sept. 9 special, “Beyond Fear: A Catholic Vision for Artificial Intelligence.”

Pope Leo XIV framed the intensifying debate over AI with the May 25 launch of his first encyclical, Magnifica Humanitas. Tech leaders said AI is rapidly reshaping society and the Church, offering powerful tools for ministry while also requiring firm ethical boundaries, human‑centered leadership, and faith‑guided discernment to ensure the technology strengthens human dignity.

“There’s a good side to that exponential innovation because it helps us get to the most vulnerable people faster,” said Joel Chakra, president of Tim Tebow Foundation and venture capitalist in AI technology companies.

“It can it can really rob us of our true humanity, or it can enhance our lives. And the choice is ours,” said Arthur C. Brooks, American author and academic who wrote the foreword to Magnifica Humanitas.

Brooks said the technology can either diminish or deepen human dignity, depending on how it is used. He added that despite real concerns about job loss and disruption, he remains optimistic because humans are “unbelievably ingenious” and made to solve problems."

Google Just Ruined One of Its Most Important Tools; The Atlantic, July 31, 2026

 Matteo Wong, The Atlantic; Google Just Ruined One of Its Most Important Tools

"Before it allowed people to create fake imagery, Google Earth was a unique resource for researchers, journalists, and intelligence officials, particularly those studying armed conflict and natural disasters. Satellite and aerial imagery can be used to monitor crop failure or deforestation, the creation of mass graves, and the results of bombing campaigns. Especially in conflict zones and other areas that are hard for people to reach, satellite imagery is a go-to method for establishing some sort of ground truth. After Google Earth rolled out the AI feature, though, one investigative reporter wrote on LinkedIn that the tool could be used to easily and rapidly create images of drone strikes that look “very close to what we’ve seen in Ukraine, Russia, the Middle East, and other war-torn conflict zones.” And instead of requiring malicious actors to find a screenshot and then get around some AI model’s safeguards, Google rolled everything needed for a misinformation farm into one smooth user interface. It took me all of four seconds to generate a rendering of a Trump Hotel on the Gaza Strip."

OPENAI ROGUE INCIDENT A CALL TO ‘DO MORE’ AS FUTURE THREATS LOOM, SAYS CATHOLIC AI ETHICS EXPERT; OSV News, July 24, 2026

 Gina Christian  . OSV News; OPENAI ROGUE INCIDENT A CALL TO ‘DO MORE’ AS FUTURE THREATS LOOM, SAYS CATHOLIC AI ETHICS EXPERT

"A recent OpenAI security incident that saw a test model go rogue may be a sign of more to come, and a call to double down on preventing future attacks, a Catholic expert on artificial intelligence ethics told OSV News.

“Cases like these are going to keep popping up, and each one of those should motivate us to do more,” said Brian Patrick Green, director of technology ethics at the Markkula Center for Applied Ethics at Santa Clara University...

The “whole thing is very significant,” Green added, noting that “there are a lot of softer targets than Hugging Face that are likely to be hit and have bad things happen.”

He said the OpenAI incident took place, providentially, within the context of a test, and that it was “significant enough for people to notice it.”

“It’s important that these are news stories,” Green said.

Given Pope Leo XIV’s recently released encyclical “Magnifica Humanitas,” which urged AI development to remain centered in God-given human dignity and the principles of Catholic social teaching, the faithful have a role to play in shaping this technology, Green said...

“For example, if you’re a Catholic who’s working in the security industry, then this is something you should work on,” he said. “If you’re a Catholic who’s working on AI models, then this is something that you should be aware of. If you’re a Catholic who’s in charge of protecting something, then it becomes particularly important for you to have adequate AI models on your side.”

Even those who are not directly involved in AI development and implementation can help, he said.

“Contact your congressperson and say, ‘Hey, I want better cybersecurity,'” said Green. “We’re super reliant on computers, and all of those systems are reliant on being secure. And if all of a sudden you make everything unsecure, then there are so many different kinds of disasters that could happen.”"

Anthropic’s Claude Fable 5 is a version of Mythos the public can access today; TechCrunch, June 9, 2026

Rebecca Bellan , TechCrunch; Anthropic’s Claude Fable 5 is a version of Mythos the public can access today

"Anthropic is bringing its most powerful AI model to the general public for the first time, but it’s doing it with guardrails. 

On Tuesday, the AI firm launched Claude Fable 5, the first publicly available version of its Mythos model. Anthropic says Fable 5 excels at software engineering, knowledge work, and vision, but it comes with hard safety limits. In high-risk areas like cybersecurity, biology, chemistry, and distillation, the model blocks responses and falls back to Claude Opus 4.8.

Pricing for both Fable 5 and Mythos 5 is $10 per million input tokens and $50 per million output tokens, double the price of Opus 4.8. That price alone might serve as a deterrent for widespread use.

Many enterprises are growing critical of AI costs after seeing the bills come inor blowing through their yearly AI budgets early. Advanced models like Opus 4.8 can exacerbate those issues, with advanced reasoning skills that can split a single request into multiple tasks."

‘It’s a hurricane warning’: Guardrails around powerful AI models may be too late; Politico, June 7, 2026

 DANA NICKEL and  MAGGIE MILLER, Politico; ‘It’s a hurricane warning’: Guardrails around powerful AI models may be too late

The U.S. has at most six to 12 months before Beijing can compete with this new wave of hyper-advanced AI models.

"The U.S. is scrambling to strengthen guardrails around increasingly powerful artificial intelligence models before China can catch up.

It may already be running out of time.

New AI models, such as Anthropic’s Claude Mythos and OpenAI’s GPT 5.5-Cyber, have advanced faster than legislation regulating the technology can keep pace. They have both shown a remarkable ability to identify software vulnerabilities and launch cyberattacks — skills that hackers and cyber adversaries are hungry to exploit.

Recent estimates suggest that the U.S. has at most six to 12 months before Beijing gains access to a frontier model with prowess comparable to Mythos or GPT 5.5-Cyber or develops an AI competitor that could eventually be wielded as a cyber weapon...

This race to develop defensive tools against a potential barrage of AI-powered cyberattacks has been accelerated by accusations that China is stealing U.S. technologies to create copycat versions of advanced AI models via distillation attacks, by which attackers use a “teacher” model’s outputs to train their own “student” models...

As this watershed moment for AI fast approaches, the U.S. government is weighing how to support the continued development of American-made technology while balancing the need for greater guardrails.

The Trump administration has largely taken a hands-off approach to regulating the release of frontier models to avoid stifling innovation and to stay competitive with China. It was finally motivated to act after Anthropic warned that the rate of AI progress threatened to upend global economies, public safety and national security if not deployed safely.

President Donald Trump signed an executive order earlier this week that encourages AI companies to submit their powerful new models for voluntary government review at least 30 days before releasing them to the public."

Poll: The midterms' new big players are pushing agendas that voters don’t fully support; Politico, May 3, 2026

ERIN DOHERTY,  JASPER GOODMAN, JESSICA PIPER,  DANIEL BARNES and BRENDAN BORDELON, Politico ; Poll: The midterms' new big players are pushing agendas that voters don’t fully support

"Deep-pocketed political groups tied to artificial intelligence and cryptocurrency are rapidly reshaping the midterm money landscape — but many Americans are uneasy with the industries behind the spending.

New results from The POLITICO Poll find broad public skepticism about crypto and AI, creating a possible conflict for candidates benefitting from an influx of contributions from the two industries. These groups are pouring millions of dollars into competitive 2026 races to elevate politicians who they believe will support their agendas in Washington.

Meanwhile, Americans have been slow to embrace either technology.

A 45 percent plurality of Americans say investing in cryptocurrency is not worth the risk, even if it can yield high returns, and a 44 percent plurality say AI is developing too quickly, according to the April survey conducted by independent firm Public First.

Nearly half of Americans say they trust a traditional bank with their money more than a cryptocurrency platform, while just 17 percent say the opposite. And two-thirds support lawmakers either imposing strict regulations or setting broad principles for the AI industry."

Nobody is governing AI; Quartz, April 8, 2026

Jackie Snow, Quartz ; Nobody is governing AI

Artificial intelligence is advancing faster than lawmakers can regulate it, while global AI governance fragments in real time

"Artificial intelligence is now making hiring decisions, tutoring children, optimizing power grids, and targeting weapons systems. The rules governing any of that are, almost everywhere, either nonexistent, stalled in committee, or under active attack.

In the United States, the federal government has spent three years producing executive orders, frameworks, and guidelines, none of which have become law. States that tried to fill the gap have been threatened with funding cuts and lawsuits. In Europe, the most ambitious AI legislation in the world is being delayed or softened before most of it has even taken effect. The technology, meanwhile, has not paused for any of this."

Claude's Constitution; Anthropic, January 21, 2026

 Anthropic, Claude's Constitution

Our vision for Claude's character

"Claude’s constitution is a detailed description of Anthropic’s intentions for Claude’s values and behavior. It plays a crucial role in our training process, and its content directly shapes Claude’s behavior. It’s also the final authority on our vision for Claude, and our aim is for all of our other guidance and training to be consistent with it.

Training models is a difficult task, and Claude’s behavior might not always reflect the constitution’s ideals. We will be open—for example, in our system cards—about the ways in which Claude’s behavior comes apart from our intentions. But we think transparency about those intentions is important regardless.

The document is written with Claude as its primary audience, so it might read differently than you’d expect. For example, it’s optimized for precision over accessibility, and it covers various topics that may be of less interest to human readers. We also discuss Claude in terms normally reserved for humans (e.g., “virtue,” “wisdom”). We do this because we expect Claude’s reasoning to draw on human concepts by default, given the role of human text in Claude’s training; and we think encouraging Claude to embrace certain human-like qualities may be actively desirable.

This constitution is written for our mainline, general-access Claude models. We have some models built for specialized uses that don’t fully fit this constitution; as we continue to develop products for specialized use cases, we will continue to evaluate how to best ensure our models meet the core objectives outlined in this constitution.

For a summary of the constitution, and for more discussion of how we’re thinking about it, see our blog post “Claude’s new constitution.”

Powerful AI models will be a new kind of force in the world, and people creating them have a chance to help them embody the best in humanity. We hope this constitution is a step in that direction.

We’re releasing Claude’s constitution in full under a Creative Commons CC0 1.0 Deed, meaning it can be freely used by anyone for any purpose without asking for permission.

Many people at Anthropic and beyond contributed to the creation of this document, as did several Claude models. Amanda Askell is the primary author and wrote the majority of the text. Joe Carlsmith wrote significant parts of many sections and played a core role in revising the text. Chris Olah, Jared Kaplan, and Holden Karnofsky made significant contributions to its content and development. More detailed contribution statement and acknowledgments below.

The preface and the acknowledgements are not part of the official constitution."

AI Is a Threat to Everything the American People Hold Dear.; The Wall Street Journal, April 2, 2026

 Bernie Sanders , The Wall Street Journal; AI Is a Threat to Everything the American People Hold Dear. It kills jobs, equality, connection, democracy and maybe the human race. Congress must act.

"The American people are deeply apprehensive about the impact that artificial intelligence will have on their lives. A recent Quinnipiac poll found that 55% of Americans think AI will do more harm than good, 70% think AI will lead to fewer jobs, and only 5% think AI development is being led by people and organizations that represent their interests.

In the midst of all of this deep concern about the future of AI, 74% of Americans think the government isn't doing enough to regulate the use of AI."

New Political Group to Push Trump’s A.I. Agenda in Midterms; The New York Times, March 29, 2026

Theodore Schleifer, The New York Times; New Political Group to Push Trump’s A.I. Agenda in Midterms 

"A new political operation with strong ties to the Trump administration is preparing to spend big money to boost President Trump’s record on artificial intelligence.

The group, called Innovation Council Action, said on Sunday that it would spend at least $100 million this year on its activities. That will include a major advocacy push behind new A.I. policy guidelines unveiled by the White House this month that seek to block state laws regulating A.I. The group is organized as a nonprofit, but is likely to start a super PAC as part of that $100 million push. That structure would allow Innovation Council to help backers and attack opponents of Mr. Trump’s A.I. agenda...

Innovation Council, by contrast, is explicitly aligned with the Trump operation. It is led by Taylor Budowich, a longtime Trump political adviser who served as White House deputy chief of staff, and has the blessing of David Sacks, a White House official."