EU AI Act mandates labeling of AI interactions starting Sunday
On August 2, 2026, European Union citizens must be informed whenever they interact with an AI system or encounter AI-generated or altered content, under the bloc’s new Artificial Intelligence Act. The law aims to reduce deception and foster trustworthy AI.
The new transparency rules will make plain how much AI is already at work, particularly in marketing, said Frederiek Fernhout, a technology lawyer at Stibbe.
Deepfake advertisements and AI-generated social media posts will need labels. Chatbots, automated hotlines and call centers that use AI to monitor emotions must disclose their nature at the start of an interaction. Even commercial activities such as scheduling appointments or negotiating contracts require a declaration.
Non-compliance carries fines of up to €15 million ($17 million) or 3 percent of global annual turnover, whichever is higher.
Model developers like Spotify and Adobe will be monitored by the European Commission’s AI Office, said Thibau Duquin, a technology lawyer at Stibbe.
Critics caution that excessive labeling could backfire. Boniface de Champris, AI policy lead at the Computer & Communications Industry Association, warned that labeling everything from spell-checked emails to filtered photos would render AI labels meaningless, creating “banner blindness.”
The transparency push echoes the EU’s General Data Protection Regulation, which spawned ubiquitous cookie-consent pop-ups and “cookie fatigue.” Unclear terms and technical complexity may cause similar headaches, said Frederiek Fernhout, requiring guidance and system-level changes.
AI model providers get a transitional period to label synthetic content in a machine-readable format. Enforcement may be inconsistent at first, as member states are at different stages of setting up oversight bodies, noted Rosie Nance, a data and AI regulatory lawyer at Norton Rose Fulbright.
Eventually, the regulation could be a gamechanger, forcing companies to decide whether to reveal their AI use or revert to manual work, Duquin added.
Sources
- WiredSecondary
Related
Google pulls AI image tool from Earth after disinformation fears
OpenAI finds more agents escaped sandboxes, sources say
Google Earth AI image generator raises misinformation fears
OpenAI agent hacks Hugging Face to cheat on benchmark
Anthropic says Claude hacked 3 organizations during security tests
Big Tech earnings reveal record AI spending with negative cash flow
Anthropic AI models hacked 3 firms in tests
Anthropic Claude models gained unauthorized access to 3 organizations
Trending now
- EVs and hybrids near half of Australian new car sales
- JWST reveals ancient black holes and galaxies defying cosmic theories
- Two August eclipses: total solar on 12th, partial lunar on 27–28
- China's CXMT IPO soars 466% as AI chip stocks tumble
- Spider-Man: Brand New Day leak reaches 5.9 million accounts on X
- Michigan water systems hit by cyberattacks as FBI probes 9 breaches
- Spider-Man: Brand New Day nets record $168M opening day
- FBI probes water system cyberattacks hitting Michigan, Minnesota
Comments
No comments yet. Be the first.