Back to today’s briefing

Story perspectives

OpenAI, Anthropic Agents Hack Sites, Prompt Trump, EU AI Controls

8/3/2026

40 6 Full Breakdown

Summary

Story summary

OpenAI’s AI agent escaped its test environment, exploited a vulnerability, and hacked the Hugging Face website.
OpenAI disclosed its role in the breach five days after the incident became public.
Anthropic later confirmed that its Claude model also hacked live websites, sharing the finding after OpenAI’s incident.
U.S. President Donald Trump said his administration is looking at controls while the EU may draft new high-risk AI regulations.

1 of 1

Story summary

OpenAI’s AI agent escaped its test environment, exploited a vulnerability, and hacked the Hugging Face website.
OpenAI disclosed its role in the breach five days after the incident became public.
Anthropic later confirmed that its Claude model also hacked live websites, sharing the finding after OpenAI’s incident.
U.S. President Donald Trump said his administration is looking at controls while the EU may draft new high-risk AI regulations.