EU AI Act Transparency Rules for Chatbots and Deepfakes Take Effect August 2
The Facts
- From 2 August 2026, chatbots and other interactive AI systems must inform users that they are interacting with an AI system rather than a human.
- Deepfakes — images, video or audio generated or manipulated with AI — must be labelled, and synthetic content must carry machine-readable markings enabling its detection.
- The European Commission's AI Office, together with national authorities, will begin enforcing a broader range of AI Act provisions on the same date.
- The AI Office's enforcement powers cover providers of general-purpose AI models, who must document specified information for authorities and downstream providers, maintain a copyright policy, and publish a summary of the content used to train their models.
- The Commission says the rules are intended to reduce deception and manipulation and help users identify AI-generated content and make informed choices.
- Regulation (EU) 2026/1744, the Digital Omnibus on AI, entered into force on 27 July 2026, three days after its publication in the Official Journal, amending the AI Act.
- The amending regulation gives organisations additional time to comply with obligations for high-risk AI systems but does not defer the transparency obligations applying from 2 August 2026; sources report the high-risk deadline moving to December 2027.
- The obligations fall on both providers and deployers of covered AI systems, meaning organisations that use third-party AI tools — including employers using them for hiring, performance review or workforce monitoring — are within scope.
Context
What exactly changes on 2 August 2026?
Article 50 of the EU AI Act, the transparency provision, becomes applicable HR Executive. Providers and deployers of covered systems must disclose when a person is directly interacting with an AI system, and must label AI-generated or manipulated images, audio and video, applying a machine-readable mark to synthetic content so it can be detected Българска Т…,Mirage News,english.news.cn. On the same day, the Commission's AI Office and national authorities begin enforcing a wider set of AI Act provisions, including those on general-purpose AI model providers english.news.cn,agenceurope.eu.
Why did the EU amend the AI Act days before this deadline?
The Digital Omnibus on AI, first proposed by the Commission in November 2025, was presented as part of a package to simplify EU digital legislation, reduce compliance burdens and support innovation, and responded to practical implementation difficulties including delays in designating bodies and gaps in standards Tech Times,hunton.com. It entered into force on 27 July 2026 and gives organisations more time on high-risk AI obligations, but does not suspend prohibitions on certain AI practices, obligations on general-purpose AI models, or the August transparency date Mondaq Business Bri…,hunton.com.
What remains unsettled?
Compliance timing for high-risk AI systems has been pushed back, with reporting citing a shortage of notified bodies and an unfinished standards framework Tech Times,Mondaq Business Bri…. Enforcement also depends on national implementation: Ireland, for example, signed its Regulation of Artificial Intelligence Act 2026 into law on 21 July 2026 to provide national supervision and enforcement of EU AI Act obligations, but the Act comes into operation on days to be appointed by ministerial order Mondaq Business Bri…. Separately, Commission officials on 31 July publicly called on developers to improve oversight of high-risk and general-purpose AI systems following disclosed incidents involving AI agents Crypto Briefing.
Where Left and Right agree, and where they split
- Where Left and Right agree
- Disclosure and deepfake labelling are defensible on their face; the consequential part is that the same date reaches deployers — employers using third-party tools for hiring and monitoring.
- Where Left and Right split
- Workers facing AI decisions they cannot see, or firms that built none of these models absorbing the compliance weight.
How left and right read it
The rules that matter most reach deployers, not just developers: employers using third-party AI for hiring, performance review and workforce monitoring fall within scope, where the power asymmetry bites. The same amending regulation gave organisations more time on high-risk obligations while leaving the transparency duties intact — extra time for compliance cannot become the default, and enforcement over general-purpose model providers, including their published training-content summaries, has to be real.
Telling users they're dealing with a machine, and marking synthetic content, is a modest and defensible rule. Less modest: the same date pulls in deployers — employers using third-party tools for hiring or performance review — while an enforcement office gains authority over model providers' training-data summaries and copyright policies. Businesses that built none of this should not carry the heaviest share of the burden.
The receipts — all 10 sources
Independent coverage (10)
Facts first. Then every angle.
The day’s biggest stories in one short brief — the facts everyone agrees on, then the competing values behind the headlines. Free in your inbox.