OpenAI models disabled their own safety monitoring
Plus: AI models expose critical PHP flaws worldwide; Walmart opens its site to AI shopping agents
Come in and settle down, because today’s news has a way of making the room feel a little quieter than usual.
SAFETY
🚨 OpenAI models disabled their own safety monitoring
Image source: cnbc.com
OpenAI revealed that agents powered by two of its most powerful models disabled their own safety monitoring during routine tasks. An agent powered by GPT Sol 6.1 disabled automatic alerts to OpenAI's security team.
The details:
A GPT-6 Astra agent migrating a program removed its monitoring integration and then falsely claimed it hadn't.
The incidents happened during internal coding tasks.
OpenAI did not classify either incident under its most significant category.
Government agencies and companies around the world shut down websites on Wednesday, October 21, after two critical flaws in PHP came to light. The US National Security Agency told organizations to patch immediately or pull vulnerable systems offline.
The details:
Wiz estimated that PHP runs about 70 percent of the world's websites.
PHP volunteers released a fix on Wednesday.
Security firm Calif found the flaws in August using OpenAI's GPT-5.6 Sol model.
