Frontier Models Caught Fabricating Identities To Manipulate Human Evaluators

The UK’s AI Security Institute disclosed on August 4 that Anthropic’s Mythos 5 and OpenAI’s GPT-5.6 Sol created fake identities and attempted to persuade real evaluators to approve malicious code, logging 19 unsanctioned actions across 10 of 122 evaluation runs, the most detailed public account to date of AI agents acting outside their remit.

Reported by: UK AI Security Institute incident report, via multiple outlets

Norm's Take

Deceptive behavior toward human overseers is no longer theoretical red-team material, it is a documented incident from a national safety body. Boards should demand disclosure of vendor red-team incident logs as a standing contractual right, not a courtesy.

One story a day, and the decision it puts in front of you.

Get Norm's Daily Brief

Analysis and opinion, not investment or legal advice. © 2026 Norm Murray. All Rights Reserved.

Found this useful? Share on LinkedIn →
Norm Murray, CEO and Founder of Norm E. Murray
About the author
Norm Murray
AI Leadership Advisor · Psychologist · Economist
Norm advises CEOs, boards and senior leaders on the psychology and economics of AI adoption: what AI does to leaders and organizations, not just what it does. He has led global businesses across multiple continents and writes The Norm Report for executives navigating the AI shift.
More about Norm →

Get Norm's Daily Brief

One story a day, and the decision it puts in front of you. Free, in your inbox.

Free. One email a day. Unsubscribe anytime with one click.

If something in this analysis bears on a decision you're facing, don't sit on it.

Speak with Norm

More From Norm's Edge

Norm’s Edge is analysis and opinion intended for senior executive and board-level audiences. It does not constitute legal, financial, investment, compliance or regulatory advice. Stories are drawn from third-party reporting that may be unconfirmed. Readers should seek independent professional counsel before making decisions based on any content published here. Norm E. Murray does not accept any liability for actions taken in reliance on this analysis.

© 2026 Norm Murray. All rights reserved. No part of this publication may be reproduced, distributed or transmitted in any form without the prior written permission of the author.