AUTOMATED SUMMARY

Government investigators bypassed safety protections on four major AI chatbots during undercover testing this spring, according to a Government Accountability Office report released Wednesday. The watchdog says each platform produced material that could help plan attacks on crowded locations or the electrical grid.

The companies say they are continuing to improve their safeguards, and some dispute how the testing was run. Lawmakers plan hearings on AI safety when the new Congress convenes in January.

FACTS

01   Investigators consulted experts and ran real-world checks, judging all four platforms' output feasible enough to help a planner. FOX NEWS

Facts behind the paywall: “the chatbot-generated parking garage image and its bomb-placement recommendation, plus GAO's undisclosed bypass methods”

The fact-sheet below is for members only.

Instead of going through each source and separating the facts from the opinions and agendas, become a member and get the fact sheet. Same story, five minutes, no spin to unpick.

Get 30 days for $1.99 →

Already a subscriber? Sign in

A subscription gets you:

  • ⚡ The 5 minute morning briefing, Monday–Friday
  • ⚡ Fast, clear summaries of stories that matter
  • ⚡ Day’s biggest story with uncensored facts and insights
  • ⚡ Fact-checks on viral hoaxes
  • ⚡ Spotlight on stories the media is burying
  • 🎧 Audio version for on-the-go

See more Upward News in Google: add us as a preferred source