Deepfakes, ChatGPT harm and audit trails that actually work
Four stories this morning on what happens when AI meets the real world, from false death reports to authentication in court.
- 01
False report claims Patrick Stewart died
A false news report circulated claiming British actor Patrick Stewart died at 86. Appears to be AI-generated misinformation or a deepfake.
Why it matters for evidence: Under Article 50(4), whoever made it must say it's AI-generated or manipulated, unless it's clearly satire or fiction.
Source: heise online (in German)
- 02
Man hospitalised after ChatGPT medical advice
A 39-year-old in San Francisco was hospitalised after following incorrect medication advice from ChatGPT. Illustrates both product liability for AI guidance and the challenge of proving what the system recommended.
Why it matters for evidence: The evidence question: could the provider show what ChatGPT actually said, and which version was running?
Source: OECD AI Incidents Monitor
- 03
What an AI audit trail should contain
Jama Software explains what belongs in an AI audit trail: which questions the AI received, what it produced, which model was used, and who approved it.
Why it matters for evidence: Article 12 requires high-risk AI to log events automatically. The hard part is keeping those logs trustworthy.
Source: jamasoftware.com
- 04
Forensics firm authenticates AI evidence for court
A consultancy now offers authentication services for AI-generated evidence, examining contested photos, video, audio and documents to provide counsel with written findings on authenticity.
Source: e3intel.io
- 05
AI Act enforcement begins from 2 August 2026
From 2 August 2026, the EU AI Act moves from preparation to enforcement. Companies using AI must now understand which specific requirements apply to their systems.
Source: instagram.com (in German)
- 06
European AI Office established for enforcement
The European Commission has established the European AI Office to support implementation of the AI Act. The Office will prepare guidance, monitor compliance, and enforce rules for general-purpose AI.
Source: digital-strategy.ec.europa.eu