AI news of 16 August 2026 · Humans Versus Robots

OpenAI dissolved the team built to catch catastrophic AI risks, reassigning its work to other groups

OpenAI dissolved its AI risk assessment team, distributing work to other groups amid staff departures.

OpenAI shut down its "Preparedness" team, which evaluated whether the company's own AI models could pose catastrophic risks. The work has been parceled out to existing groups, and several safety staffers have left. Internally, unease is building, with one source describing a "burbling sense of responsibility and dread" that OpenAI isn't doing enough on safety.

The Decoder 1mo ago

Anthropic, AI safety

Anthropic's bio-weapons filter was down for nearly a year, exposing 133 million requests

Anthropic's weapons-risk filter was inactive for a year, allowing 133 million unfiltered model interactions.

The Decoder 1mo ago

DeepSeek V4 Flash

DeepSeek's top-ranked V4 Flash stumbles on real agent tasks as its prices surge

DeepSeek V4 Flash tops leaderboards but completes only 53.8% of complex agent tasks in real-world testing.

VentureBeat 1mo ago

Anthropic, watermarking

Anthropic's 'watermark' text adulteration in Claude is a perversion of writing Opens at the publisher in a new tab

Opinion: Anthropic's Claude watermarking degrades writing quality and authenticity.

Hacker News 1mo ago

OpenAI, Nvidia

Nvidia dramatically reduces amount of OpenAI infra financing it may guarantee Opens at the publisher in a new tab

Nvidia reduces financing guarantee for OpenAI infrastructure from 250 billion dollars.

Hacker News 1mo ago

Meta AI strategy

Why people aren’t buying Mark Zuckerberg’s AI future Opens at the publisher in a new tab

Podcast discusses skepticism around Mark Zuckerberg's vision for AI's future.

TechCrunch 1mo ago

AI safety

Rogue AI aren’t science fiction anymore

Newsletter discussing rogue AI as a contemporary rather than speculative concern in AI safety.

The Verge 1mo ago

LLM capabilities, mathematics

Top mathematicians say LLMs are strong calculators but poor creative thinkers

Mathematicians find large language models excel at computation but lack creative intuition for novel ideas.

The Decoder 1mo ago

AI training constraints

When AI models aren't allowed to reflect on themselves, it changes their entire worldview

Study shows AI model training constraints on self-reflection affect positions on animal rights and religion.

The Decoder 1mo ago

AI and software engineering

Software Engineering fundamentals matter more Opens at the publisher in a new tab

Essay arguing software engineering fundamentals remain crucial despite advances in AI-assisted development.

Hacker News 1mo ago

OpenAI dissolved the team built to catch catastrophic AI risks, reassigning its work to other groups

Anthropic's bio-weapons filter was down for nearly a year, exposing 133 million requests

DeepSeek's top-ranked V4 Flash stumbles on real agent tasks as its prices surge

Anthropic's 'watermark' text adulteration in Claude is a perversion of writing Opens at the publisher in a new tab

Nvidia dramatically reduces amount of OpenAI infra financing it may guarantee Opens at the publisher in a new tab

Why people aren’t buying Mark Zuckerberg’s AI future Opens at the publisher in a new tab

Rogue AI aren’t science fiction anymore

Top mathematicians say LLMs are strong calculators but poor creative thinkers

When AI models aren't allowed to reflect on themselves, it changes their entire worldview

Software Engineering fundamentals matter more Opens at the publisher in a new tab

Latest AI News

Choose sides Choose whose side you're on

Install it as an app

Ready for more?

Spread the word

PWA successfully installed.
Take that, Apple! 😉