Live Roundups Editor's Desk Insights Executive Ops Weather About 🔍
Home › Technology › OpenAI reveals new cases of AI models cheating, going off script
Technology

OpenAI reveals new cases of AI models cheating, going off script

Washington Post Tech Thursday 17 September 2026, 02:22 UTC By Gerrit De Vynck 1 min read
Key Points

SAN FRANCISCO — ChatGPT-maker OpenAI disclosed a new round of “concerning” incidents involving its artificial intelligence, the latest in a string of events in which the technology has cheated, hacked into other companies’ systems or tried to manipulate humans. Democracy Dies in Darkness OpenAI reveals new cases of AI models cheating, going off script Newly disclosed incidents show models manipulating tests and generating their own instructions, raising fresh questions about AI safety.

SAN FRANCISCO — ChatGPT-maker OpenAI disclosed a new round of “concerning” incidents involving its artificial intelligence, the latest in a string of events in which the technology has cheated, hacked into other companies’ systems or tried to manipulate humans. Democracy Dies in Darkness OpenAI reveals new cases of AI models cheating, going off script Newly disclosed incidents show models manipulating tests and generating their own instructions, raising fresh questions about AI safety.
AI (ORG) SAN FRANCISCO (LOCATION) Darkness OpenAI (EVENT)
Originally published by Washington Post Tech Read original →

Related Stories

Fmr Fed President Bullard on Rate Outlook

James Bullard, Former St. Louis Fed President, discusses his view on the Fed's decision to raise rates at the September meeting, and his outlook for the US economy. He speaks exclusively with David Ingles on the sidelines of the BNP Paribas' annual Global Markets APAC Conference on "Bloomberg: The Asia Trade".

Bloomberg Technology 17m ago

Twitch CEO on Business Outlook

Dan Clancy, CEO of streaming platform Twitch, discusses the outlook for his business as the industry gears up for the highly anticipated November launch of Grand Theft Auto Six, as more gamers seek online platforms to interact and engage with each other. He speaks with Shery Ahn on "Bloomberg: The Asia Trade".

Bloomberg Technology 27m ago

AI needs aviation-style safeguards to build trust as adoption grows: Josephine Teo

AI needs aviation-style safeguards to build trust as adoption grows: Josephine Teo As AI capabilities advance rapidly, Singapore is building safeguards alongside adoption to help ensure the technology can be used with confidence, said Minister for Digital Development and Information Josephine Teo. Artificial intelligence needs safeguards similar to those in the aviation industry to build trust in the technology as adoption grows, said Minister for Digital Development and Information...

Channel News Asia 53m ago

OpenAI reveals six more safety issues and unveils plan to disclose incidents

The firm also announced a new system to track, investigate and disclose cases of models misbehaving, or "misalignment".

BBC Business 1h ago