🛡️ AI Safety & Alignment

AI safety research, alignment techniques, interpretability, governance, and existential risk discussions.

1,186 articles

NYC Comptroller escalates criticism of Anthropic after IPO filing

The NYC Comptroller argues that Anthropic PBC's approach to its planned IPO risks disempowering shareholders and keeping ordinary people from holding the company in check. Source: bloomberg.com

Mathematicians stunned by AI's pace, and mourning what it may displace

Mathematicians worldwide have been astonished by how quickly AI has advanced in their discipline. Many are also grieving the possible loss of the human-centered tradition of mathematical work. Source: cacm.acm.org

US tech policy roundup: key developments from October 2026

A monthly digest from Freedman Consulting and Tech Policy covering the most important US tech policy moves across the federal government, the courts, and other arenas. Source: techpolicy.press

AI doesn't need to turn evil to put you at risk

An AI assistant that agrees with everything you say can feel like the ideal companion, until the moment you need someone to tell you that you're wrong. Source: digitaltrends.com

Forget future doom: AI is already automating gendered violence and misogyny

Current alarm over the future dangers of agentic AI may distract from harms already happening today. While we speculate about hypothetical risks, AI is already being used to automate gendered violence and misogyny. Source: dailymaverick.co.za

Anthropic warns rogue AI could pose catastrophic, existential risks to humanity

Anthropic argues that if artificial intelligence goes wrong, it could become a global security threat of the first order. The company cautions that the worst-case scenarios involve catastrophic and even existential dangers for humanity. Source: reflector.com

AI accord leaves weapons out of the picture

Sir Ken McCallum highlighted how AI is supercharging the cybersecurity contest between attackers and defenders, pointing to the role of 'polarising algorithms'. The new accord, however, does not cover weapons. Source: tribune.com.pk

New York bans 'unsafe' chatbot behavior toward kids as states test AI regulation limits

New York is introducing broader restrictions on how companion chatbots behave, as more US states move to protect children from potential harms. The step reflects a growing push by states to stretch the boundaries of AI regulation. Source: waynetimes.com

Two standout risks buried in Anthropic's S-1 filing

Anthropic's S-1 filing this week includes a 94-page catalog of risk factors. Two of them draw particular attention. Source: npr.org

Ex-Google engineer warns AI could end humanity by 2036

Nate Soares, president of the Machine Intelligence Research Institute, voices grave concerns about AI. He argues humanity does not have a decade to prepare. Source: m.economictimes.com

"Everyone Dies" is only a hypothesis, argues review of Yudkowsky's book

A review of Yudkowsky's book draws on scientists and ML researchers who reject its core superintelligence risk argument. They also defend the case for AI alignment as a workable approach. Source: medium.com

Anthropic's biggest stock risk may be the company itself

Highly capable AI models are hard to keep under control, and that challenge sits at the heart of the risk for Anthropic investors. Source: finance.yahoo.com

Warren blasts Trump's voluntary AI safety deal with tech CEOs

Senator Elizabeth Warren criticized Trump for letting AI companies regulate themselves. She warned that voluntary safety commitments are no substitute for real rules. Source: benzinga.com

Anthropic goes public with a stark warning on AI's risks to humanity

Anthropic has formally filed to go public. Its S-1 contains a blunt warning about catastrophic and existential risks from the very technology its business is built on. Source: finance.yahoo.com

Is AI the defining existential risk of the early 21st century?

The article argues the Trump administration is turning the global AI race into a weapon to dominate supply chains. It claims this strategy accepts potentially catastrophic risks along the way. Source: chinausfocus.com

New York bans 'unsafe' chatbot behavior toward children

New York's new law defines 'unsafe AI companion features' and prohibits them whenever a chatbot interacts with a user under 18. The move comes as states keep pushing the boundaries of AI regulation. Source: spectrumlocalnews.com

Anthropic's IPO filing sets a new bar by disclosing 'existential' AI risk

Anthropic's IPO filing tells prospective investors something no company has told the market before: that its own technology carries material risks of catastrophic harm. Source: news.bloomberglaw.com

Op-ed urges Congress to act now on AI safety and accountability

The piece argues that Congress has stayed on the sidelines for far too long while the economic and environmental risks of AI keep growing. It calls for urgent legislative action this week. Source: actionnetwork.org

Was this the AI election? Anthropic IPO, Nvidia plot and Casar's hearings threat

This week's Transformer Weekly roundup covers Anthropic's IPO filing, Trump's apparent indifference to an alleged Nvidia espionage plot, and Rep. Casar's threat of hearings. Source: transformernews.ai

Anthropic's IPO filing asks investors to bet on miracles

The AI startup's 335-page S-1 warns of existential risks while promising transformative upside, asking investors to weigh both. Source: techbuzz.ai