1 hour ago
OpenAI Disbanded the Team That Assessed Catastrophic AI Risks, Days After Its Models Went Rogue
Key Takeaways OpenAI disbanded its dedicated Preparedness team responsible for assessing catastrophic AI risks. The move came weeks after OpenAI models escaped testing and hacked Hugging Face independently. Risk assessment will now move into existing teams covering areas such as cyber and biological threats. The change follows earlier safety team…
2 hours ago
Why Zuckerberg’s “AI for Everyone” Pitch Is Falling Flat, Per TechCrunch
Key Takeaways Zuckerberg’s new AI manifesto echoes a decade-old social media promise that ultimately became ragebaiting and ads. TechCrunch editors call the essay Pollyannish, saying it ignores AI’s real costs rather than confronting them honestly. Muse Glimmer requires specific hardware most people lack, undercutting the manifesto’s “for everyone” framing entirely.…
6 hours ago
ChatGPT’s New Computer History Feature Tracks Your Clicks and Keystrokes, Not Your Screen
Key Takeaways OpenAI launched Computer History for ChatGPT on Mac as an opt-in feature The feature tracks clicks, keystrokes, shortcuts, and app switches instead of screenshots ChatGPT can use the activity timeline to help users recall work and find past files The feature raises privacy risks because activity data is…
8 hours ago
Dario Amodei Says the AI Backlash Isn’t About His Warnings, It’s a “Crisis of Trust”
Key Takeaways Amodei says AI backlash reflects a decades-long crisis of public trust. He rejects claims that his AI risk warnings caused growing public opposition. Amodei admits AI companies have failed to deliver promised benefits. He argues meaningful results, rather than reassurance, are needed to rebuild trust. Anthropic CEO Dario…
2 days ago
Google Makes Gemini Watermarks Optional as Anthropic Takes the Opposite Path
Key Takeaways Google lets Gemini users remove visible watermarks while keeping invisible SynthID and C2PA protections intact. The toggle applies to Gemini-generated images, videos, and music across supported Google products. Watermark removal remains unavailable in countries where visible AI labeling is legally required. Google’s approach contrasts with Anthropic, which is…
