How to Cut Through AI Hype and Protect Your Privacy: A Guide from EFF

Every week, it seems, a new company announces an AI tool that will “revolutionise” your workflow, “personalise” your experience, or “intelligently” anticipate your needs. But behind the marketing language, many of these products collect more data than they disclose, rely on poorly documented training methods, and rarely deliver on their boldest promises. The result is a fog of hype that makes it hard for ordinary users to know which tools deserve their trust – and which ones put their privacy at risk.

The Electronic Frontier Foundation (EFF) has been working since 1990 to defend digital rights. Over the past few years, the organisation has published a series of reports and articles that cut through the noise, calling for rational regulation of AI and exposing the real harms behind exaggerated claims. Their message is simple: we need to evaluate AI tools on evidence, not enthusiasm.

What Happened

In recent months, EFF has released several important pieces that highlight the gap between AI marketing and reality. Their article Help EFF Cut the AI Hype urges readers to send in examples of dubious claims. Automated Moderation Is Here to Stay examines how platforms offload difficult decisions to opaque algorithms. AI Regulation Should Be Rational, Not Retaliatory argues against knee‑jerk laws that could stifle beneficial uses while failing to address real harms. Another piece, Blocking the Internet Archive Won’t Stop AI, But It Will Erase the Web’s Historical Record, shows how overreacting to AI – in this case, blocking the Internet Archive – can destroy valuable public resources without meaningfully curbing misuse. EFF also released a New Report Helps Journalists Dig Deeper Into Police Surveillance Technology, equipping reporters to ask better questions about AI‑powered policing tools. And their 2023 document Generative AI Policy Must Be Precise, Careful, and Practical laid out principles for drafting legislation that protects rights without causing collateral damage.

Together, these articles form a coherent, evidence‑based critique of AI hype. EFF isn’t against AI; it’s against the abdication of responsibility that comes when companies and governments treat AI as magic.

Why It Matters

The problem with AI hype isn’t just that it wastes money or disappoints users. It has concrete privacy implications. When a company claims its AI can “personalise” your experience, it often means it is collecting vast amounts of your data – browsing history, location, biometrics, even voice recordings – without clearly explaining how the data will be used or stored. Exaggerated claims also fuel surveillance creep: police departments buy facial recognition systems based on vendor promises of near‑perfect accuracy, even though audits repeatedly show significant error rates, especially for people of colour. And when policymakers react to AI panic by passing vague, retaliatory laws, they can inadvertently ban or restrict beneficial uses while doing little to stop the most harmful ones.

The Internet Archive example is instructive. After some publishers and authors sued over use of copyrighted books for training AI models, courts ordered the Archive to block access to certain materials. EFF argued that this response punishes a non‑profit library for the actions of large AI companies. The result is a weaker historical record for everyone, while the powerful have the resources to litigate or license around the blocks.

What You Can Do

You don’t need to be a technologist to cut through the AI hype. Here are practical steps you can take before trusting an AI‑powered service with your data:

  1. Ask what data it collects. Look for a privacy policy that lists data types and retention periods. If the policy is vague or says “we collect information to improve our services,” that’s a red flag.
  2. Find out how the model was trained. Companies should disclose whether the training data includes personal information, copyrighted material, or biased sources. If they won’t say, assume the worst.
  3. Check if the tool can be audited. Independent researchers need access to test for bias, accuracy, and security. If the developer blocks outside audits, that’s a reason to be cautious.
  4. Look for specific use cases, not generic promises. A claim like “our AI helps doctors diagnose faster” is less meaningful than “our AI was tested on 5,000 X‑rays from three hospitals and detected lung nodules with 92% sensitivity.”
  5. Support organisations like EFF that do the difficult work of scrutinising AI claims. You can donate, share their articles, or simply amplify their message when you see exaggerated hype in your own feed.

Sources

  • Electronic Frontier Foundation. Help EFF Cut the AI Hype. July 2026. Link
  • Electronic Frontier Foundation. Automated Moderation Is Here to Stay. July 2026. Link
  • Electronic Frontier Foundation. AI Regulation Should Be Rational, Not Retaliatory. June 2026. Link
  • Electronic Frontier Foundation. Blocking the Internet Archive Won’t Stop AI, But It Will Erase the Web’s Historical Record. March 2026. Link
  • Electronic Frontier Foundation. New Report Helps Journalists Dig Deeper Into Police Surveillance Technology. February 2026. Link
  • Electronic Frontier Foundation. Generative AI Policy Must Be Precise, Careful, and Practical: How to Cut Through the Hype and Spot Potential Risks in New Legislation. July 2023. Link