Short answer

You cannot reliably spot AI-generated misinformation by how it looks or reads, so check where it came from. Trace the claim to a primary source, look for the same report from independent outlets and be cautious of content that triggers a strong emotion. Technical labelling and provenance tools are being developed by NIST and required in the EU, but they are not yet a dependable filter.

Key takeaways

  • Fluent writing and realistic images are no longer evidence that something is real.
  • NIST's AI 100-4 report reviews standards and tools for authenticating content, tracking provenance and labelling synthetic content such as with watermarking.
  • The EU AI Act's Article 50, applicable from 2 August 2026, requires providers to mark synthetic output in a machine-readable way and deployers to disclose deepfakes.
  • Labels depend on the maker's cooperation, so unlabelled content is not proof of authenticity.
  • Checking the source, the date and independent confirmation beats trying to spot visual glitches.

Why looks are not evidence

A few years ago, clumsy grammar and warped hands exposed many fakes. Those clues are fading. AI can now produce clean articles, plausible quotes, realistic photos and convincing audio quickly and cheaply. Our guide on how to spot AI-generated images and video lists the visual cues that still sometimes help, but treat them as hints. A clean image can still be false, and a rough one can be real.

The more dependable question is not “was this made by AI?” but “where did this come from, and who vouches for it?”

A five-step check

  1. Feel the pull. Content built to provoke outrage or fear spreads fastest. If your pulse jumps, slow down.
  2. Find the source. Search for the original statement, filing, study or full video. A screenshot of a headline is not a source.
  3. Seek independent confirmation. Genuine events are usually reported by several unrelated outlets or confirmed by an official body.
  4. Check the details. Dates, locations, names and numbers should line up. For photos and clips, a reverse image search often shows an older or different origin.
  5. Check the messenger. How old is the account? Does it post only one kind of content? Does it ask for money?

When AI tools are the source

Chatbots can state falsehoods fluently, and they sometimes cite sources that do not exist. Before you repeat a claim from an AI answer, open the link and confirm it says what is claimed. See AI hallucinations and fake citations for how this happens.

What standards bodies and regulators are doing

NIST published AI 100-4, “Reducing Risks Posed by Synthetic Content”, in November 2024. It examines existing standards, tools and practices for authenticating content and tracking its provenance, and for labelling synthetic content, such as using watermarking. It frames transparency as a set of technical approaches rather than one fix.

The EU has put some of this into law. Article 50 of the AI Act applies from 2 August 2026. It requires providers of systems that generate synthetic audio, image, video or text to mark outputs in a machine-readable format so they are detectable as artificially generated, and requires deployers to disclose deepfakes that resemble real people, objects, places or events. Text published to inform the public on matters of public interest must be disclosed as AI-generated unless it has human review or editorial control. For a plain-language overview see the EU AI Act explained for consumers.

The limit is obvious: rules bind those who follow them. Absence of a label does not make something real.

How misinformation feeds fraud

False stories are often bait. A fake news page with a celebrity “endorsement” leads to an investment scheme; a fake charity appeal follows a real disaster. Our guides on investment scams and charity scams show how that works, and AI-generated fake reviews and scam websites covers the shopping version.

A habit worth building

Treat unverified content as a claim, not a fact. Wait for confirmation, share the source rather than the screenshot, and correct yourself publicly when you get it wrong. That habit works whether the fake was made by a person, a template or a model.

What you can do

  1. 01Pause before sharing anything that makes you angry, afraid or triumphant, especially if it is breaking news.
  2. 02Find the original source: the official statement, court record, dataset or full video, not a screenshot or a repost.
  3. 03Look for the same account from at least two independent, established outlets or an official body.
  4. 04Run a reverse image search and check the date, place and account history before trusting a photo or clip.
  5. 05Be sceptical of 'citations' from AI tools: open the link and confirm the page says what is claimed.
  6. 06If a fake involves a scam, a person or an election, report it to the platform and, for fraud, to the relevant authority.

Related reading: AI voice clone scams and the data breach tracker.

Frequently asked questions

Can tools detect AI-generated text or images?

Detection is imperfect, and NIST treats it as one part of a wider set of approaches that includes provenance tracking and labelling. Do not treat a detector's verdict as proof either way. Source-checking is more reliable.

What is content provenance?

Provenance means recording where a piece of content came from and how it was changed. NIST AI 100-4 reviews existing standards, tools and practices for authenticating content and tracking its provenance.

Will AI content be labelled?

In the EU, Article 50 of the AI Act requires providers of generative systems to make output detectable as artificial in a machine-readable format, and deployers to disclose deepfakes and AI-generated text on matters of public interest unless it has human editorial control. It applies from 2 August 2026. Bad actors will not comply.

How does misinformation connect to scams?

Fake news and fake endorsements are used to push investment, charity and shopping scams. A convincing story about a celebrity or crisis is often the bait.

Sources

  1. Reducing Risks Posed by Synthetic Content: An Overview of Technical Approaches to Digital Content Transparency (NIST AI 100-4), National Institute of Standards and Technology, 2024-11-20
  2. Article 50: Transparency obligations for providers and deployers of certain AI systems, EU AI Act (artificialintelligenceact.eu, unofficial text)

This article reflects public information as of 2026-11-06. AI tools and rules change quickly; we update articles when new facts are published. Spot an error? See our editorial and corrections policy.