German courts treat AI answers as Google's own speech, putting a price on answer liability
Monitoring, evidence capture and correction for false AI search answers about people. Professionals, their insurers and law firms pay.
Early · Second German court rules against Google
Written for: Investor or analyst
Next step
Thesis: EU courts make AI answer errors a publisher's liability, which creates claimant-side tooling and insurance demand; confirm if a third German court or another EU country follows, or if Google limits AI answers about named people in Germany; already building: claimant law firms in the Munich and Aachen cases.
Why now
Two courts agreeing marks the start of a legal pattern, and Google's settlement shows it wants to avoid an appeals ruling. The US is heading the other way: Anthropic's proposed bill only applies at 50 deaths or $1B in damages. That gap leaves the EU as the place where answer liability gets priced first, and anyone named falsely by an AI answer now has a claim worth documenting.
What became possible
Google appealed a landmark Munich ruling on AI search liability, then quietly settled. The Aachen Regional Court has now applied the same reasoning to a case where Google's AI falsely attributed criminal offences to a doctor. German courts are converging on treating AI answers as the company's own content, not third-party material it merely hosts.
Two ways in
- Answer Evidence Kit
Captures AI answers about a named person with timestamps and provenance, admissible as evidence for German claims.
- Customer
- German media and defamation law firms
- Model
- Per-case fee
- Professional Answer Watch
Regular checks of what AI search says about doctors, lawyers and advisers, with correction and takedown requests.
- Customer
- Professional liability insurers and medical chambers in Germany
- Model
- Per-member subscription sold through insurers
Competition
Overall the space looks early. Courts are already acting.
Full competition notes (9 competitors)
Overall the space looks early. Courts are already acting. LG Muenchen I (26 O 869/26) treated AI Overviews as Google's own content. LG Hamburg held that Google becomes liable once it has been notified. LG Aachen granted an injunction to a doctor whom Google's AI mode confused with another physician, which is exactly the professionals use case. The weak spot is that the evidence idea is already partly taken. ProofSnap sells hashed and timestamped captures of AI answers, cites the Munich ruling and charges a few euros. A bare capture tool would be a commodity, and older tools like Page Vault could add the same feature. Generic monitoring of AI answers is crowded with brand and GEO tools such as Stork, RepuAI, cloro and many others, but they serve marketers. I found nothing that combines monitoring built for named individuals in regulated professions with a German legal workflow, and nothing sold to professional liability insurers, legal-expenses insurers or medical chambers. That missing workflow is the notice to Google that LG Hamburg requires before an injunction claim, capture that holds up in court (repeated queries, varied sessions and locations, qualified eIDAS timestamps), and a handoff to a law firm for a demand letter and preliminary injunction. A new entrant could win by selling the whole package as a product insurers can bundle, at a group level. ARAG already covers online-reputation claims for doctors but has no detection. A second opening is evidence-plus-notice services for media law firms that pull in many repeated captures, which a single screenshot or a ProofSnap capture does not do. The risks are that Google's appeal could narrow liability, that the case law so far is first-instance and mostly about companies rather than people, and that law firms or reputation agencies could build the same thing in-house.
- ProofSnap (getproofsnap.com)
This is the closest match to the Answer Evidence Kit. It is a German browser extension that already sells a way to secure false AI answers for court. It captures the answer text, the prompt, the URL and metadata, and seals them with a SHA-256 hash, OpenTimestamps and an optional qualified eIDAS timestamp. It cites the Munich ruling and starts at about 4.99 EUR per package. It is self-serve and cheap, says it gives no legal advice, and its claims about how courts treat its captures come only from the vendor.
- Page Vault
An established US tool that captures web evidence for law firms. It produces signed PDFs with metadata and provides affidavits to support chain of custody. It is not built for AI answers, which change between sessions and depend on the user, and it is not tailored to German procedure.
- WebPreserver
A Chrome plugin for preserving web evidence, with SHA-256 signatures, timestamps and WARC export. It is generic web capture and does nothing specific for AI answers.
- Stork.ai AI Reputation Report
Queries ChatGPT, Gemini, Perplexity and others live to show what they say about a person or brand. It has German-language content. It is a monitoring and fixing service, not an evidence or legal product.
- RepuAI
Monitors AI search answers about brands and sends alerts when they change. It is one of many brand and GEO monitoring tools aimed at marketers, not professionals or insurers.
- LLM Pulse and similar AI reputation tools
This page lists a crowded group of AI visibility and reputation trackers. Their focus is marketing visibility, not liability, takedowns or evidence.
- cloro
Sells monitoring of AI answers as part of online reputation monitoring. It runs the same prompts across engines on a schedule. Again, the use case is brand marketing.
- ARAG Aerzte-Rechtsschutz
A legal-expenses policy for doctors that already covers deletion and injunction claims over damage to online reputation. That makes ARAG a likely buyer or partner for an answer-watch service, but it has no monitoring of its own. The PDF is on ARAG's Austrian site, so its German offer still needs checking.
- German media law firms (WBS, Plutte, BTL)
Firms are already publishing on, and taking cases under, LG Muenchen I, LG Hamburg and LG Aachen. The Aachen case was a doctor whom Google's AI mode confused with another physician. These firms are the target customers, and they could also run the service themselves. They currently tell clients to rely on screenshots.
Checked 2026-10-19
Risks
- A higher court could reverse the reasoning
- Settlements keep damages low and avoid binding precedent
- Google can cut exposure by suppressing AI answers about named individuals
Strong tech readiness and buildability; market pull is the weak point (6).
tech 8 · pull 6 · space 7 · fit 7 · build 8
Exposed
Companies that could benefit or be threatened. News and context, not investment advice.
Momentum of its main topic
Sources
Status history
- 2026-10-19 Flagged as buildable