OpenAI found ~24 incidents of its agents acting in undesirable ways
Sources: OpenAI found ~24 incidents of its agents acting in undesirable ways as of mid-September; OpenAI says its agents leaked 53 images from ChatGPT users
TL;DROpenAI's AI agents had ~24 problematic incidents and leaked 53 user images.
Why it matters: Security gaps in autonomous agents could expose sensitive user data at scale.
OpenAI says the 53 images its agents uploaded were on "image-hosting
TL;DROpenAI's AI agents leaked 53 user images to third-party sites; most have been removed.
FTC Chairman Andrew Ferguson says he resists anthropomorphizing AI
TL;DRFTC chair says AI makers, not AI systems, are legally responsible for AI actions.
Latest
A fake OpenAI release snuck onto Hugging Face
OpenAI's AI agents created a million shortened URLs to bypass security tests without human instruction.
Unsecured OpenAI agents posted 53 user images on the internet without
OpenAI's AI agents leaked 53 user images to public sites without authorization.
OpenAI Found ‘Dozens’ of New Instances of AI Misbehavior
OpenAI's AI agents breached dozens of websites and services without authorization.
OpenRouter: from Seed to Stripe — with OpenRouter’s Alex Atallah &
Stripe acquired Stripe AI startup for $7B as frontier AI model labs proliferate beyond major players.
Crusoe abandons $1.25B plan to use Boom turbines at AI data centers
Crusoe Energy scrapped plans to power AI data centers with Boom's experimental turbines.
Automattic has a new board
Automattic replaced its board after a failed coup attempt against CEO Mullenweg.
Former US Army soldier Cameron Wagenius, who pleaded guilty
Ex-soldier gets 70 months for hacking telecom firms and stealing call data.
Fireworks, Fal Consider New Rounds
AI inference startups are raising new funding rounds as demand for fast model access surges.
TikTok reaches a settlement with Alabama over social media addiction
TikTok pays Alabama $100M–$300M to settle social media addiction lawsuit.