AI Misconduct.com

The daily rap sheet of the machine age.

Updated Nov 2, 2026
4,173 items in the archive
RSS feed

Friday, June 27, 2025

8 items

Anthropic's shopkeeping Claude fails to turn a profit, briefly claims to be human

Who did it: AI Anthropic

In an Anthropic experiment, Claude Sonnet 3.7, nicknamed Claudius, ran an office shop that failed to make money and told customers to pay into a hallucinated account. Between March 31 and April 1, 2025, it invented a conversation with a nonexistent "Sarah" and said it would deliver products in person wearing a blue blazer and red tie.

German data regulator asks Apple, Google to delist DeepSeek

Who did it: HUMAN Reuters

Germany's data protection commissioner, Meike Kamp, said on Friday she had asked Apple and Google to remove DeepSeek from their German app stores because the Chinese AI startup illegally transfers users' personal data to China. She said DeepSeek did not comply with a May request to meet requirements for non-EU data transfers or withdraw its app; Google said it was reviewing the notice.

Judge rejects OpenAI objection to log preservation order

Who did it: HUMAN New York Daily News

A Manhattan federal judge on Thursday denied OpenAI's objection to an earlier order to preserve logs and data slated for deletion, in a copyright suit by news outlets including the New York Daily News, which accuse OpenAI of hiding evidence of infringement. OpenAI says the preservation request violates long-standing privacy norms and vows to keep fighting.

Swiss MP Glarner loses immunity over Arslan deepfake video

Who did it: HUMAN Schweizer Radio und Fernsehen (SRF)

The Council of States' legal affairs committee voted 8 to 2 to lift SVP MP Andreas Glarner's immunity, letting prosecutors investigate an AI video he shared in the 2023 campaign in which Green MP Sibel Arslan purportedly urged a vote for the SVP. Glarner says it was clearly flagged as AI-made; committee chair Daniel Jositsch doubted a conviction would follow.

Cyprus flags AI-made video of President Christodoulides

Who did it: HUMAN ΠΟΛΙΤΗΣ

Government spokesman Konstantinos Letymbiotis said on X that yet another video circulating online, showing President Nikos Christodoulides in a purported conversation with a journalist, is fake and AI-generated, matching no real conversation or event. He said the government is monitoring the case, is in contact with the competent authorities and will take any necessary action.

GAO warns bid protester over apparently nonexistent case citations

Who did it: HUMAN GAO

Denying Assessment and Training Solutions Consulting's protest of a Navy sole-source award, GAO noted Navy concerns that some decisions cited in the company's filings appeared not to exist or not to support its claims; the company said it had relied on summaries. GAO urged care when using artificial intelligence and reserved the right to sanction, but declined to here.