GEO within lawyers' professional conduct law: what BRAO and BORA permit

Why GEO is not new professional-law territory for law firms

When you as a lawyer first hear that ChatGPT or Google AI Overviews 'recommend' clients to law firms, the first question is almost always the same: am I even allowed to do that? The reassuring answer is: yes. Lawyers' advertising law has liberalized fundamentally since the Federal Constitutional Court's Bastille decision in 1987. Advertising is the rule today, prohibition the exception. What applies to your website and your Google profile applies at its core to AI answers too.

Generative Engine Optimization means nothing other than preparing your factual information so that language models understand it, reproduce it correctly and name you as a source. You are not inventing a new form of advertising, you are making existing, professionally permissible information machine-readable. The test standard remains Section 43b BRAO: advertising is permitted insofar as it informs factually, in form and content, about professional activity and is not directed at securing an individual instruction in a specific case.

The perspective matters: it is not the AI that advertises for you, but you who provide information the AI uses. That keeps you in the role of the factually informing lawyer - and it is exactly this role that protects you under professional conduct law.

Section 43b BRAO: the core and what it means for AI content

Section 43b BRAO sets three requirements you should remember: factual information about the form and content of your activity, no reference to a specific mandate, and no promotional exaggeration. For your GEO content that means, very practically: an FAQ text like 'How does a claim for unfair dismissal proceed?' is unproblematic, because it informs about your activity. A text like 'Call now because your neighbor is suing you right now' would be case-specific and therefore impermissible.

The crux with AI is that you don't fully control the output. A language model could sharpen your factual statements. Under professional conduct law, however, you are liable for what you publish yourself, not for every phrasing a model generates from it. So the maxim is: phrase your source texts so soberly and precisely that even a shortened rendering stays factual.

Avoid superlatives like 'the best law firm', 'guaranteed success' or 'Germany's leading traffic-law attorney' if you cannot objectively prove them. Such statements are not only promotional under Section 43b BRAO, they are rarely adopted by AI systems anyway, because they are unsubstantiated claims.

Section 6 BORA: factual information is the safe harbor

The professional code specifies in Section 6 BORA that a lawyer may inform about their service and their person, insofar as the information is factual and professionally related. This is precisely where the gold vein for GEO lies: factual, profession-related, verifiable information is exactly what language models prefer. Specialist lawyer titles, areas of activity, year of admission, languages, location, typical procedural sequences - all of that you may communicate, and it makes you quotable for AI at the same time.

Use the terminology clearly regulated under professional law correctly. You may only use 'specialist lawyer for family law' (Fachanwalt) if you have earned the title. 'Area of activity' (Tätigkeitsschwerpunkt) and 'area of interest' (Interessenschwerpunkt) are permissible under Section 7 BORA, but defined terms with a numerical limit. If you use these terms precisely, you give the AI unambiguous signals - and stay professionally clean at the same time.

A common mistake: firms write vaguely 'We advise in all areas of law'. That is borderline misleading under professional law and worthless for GEO, because it sets no clear competence signal. Concrete, honest information about what you really do regularly is better.

Client reviews and testimonials in AI answers

Language models and AI Overviews often draw on reviews from Google, Anwalt.de or Proven Expert. Reviews are permissible under professional conduct law, as long as they are genuine and not misleadingly steered. You may ask satisfied clients for a review. What you may not do: buy, fake or induce reviews with benefits, because that would be anti-competitive under the UWG (Unfair Competition Act) and violates the factuality requirement.

Pay particular attention to the topic of success fees and outcome promises. If a client writes in a review 'Mr. X won my case 100 percent', that is their opinion. It becomes problematic if you adopt such statements yourself in marketing texts and present them as a guarantee of success. For GEO the rule is: genuine, differentiated reviews with concrete context are more credible and are more likely to be rated by the AI as a trustworthy signal.

A confidentiality problem lurks here too. In references or case examples, never mention identifiable clients or facts without express consent. Section 43a BRAO binds you to confidentiality - including in promotional texts that AI later picks up.

Typical GEO questions from everyday firm life

People today ask language models things they wouldn't previously have dared ask a lawyer. 'What does a divorce cost?', 'Am I entitled to severance pay?', 'How long does a fine proceeding take?'. If your website answers such questions factually and precisely, you become the source the AI draws from. That is impeccable under professional conduct law, because you inform generally about the legal situation and your activity without initiating an individual mandate.

With cost questions, caution is called for. You may inform generally about the RVG (lawyers' fee schedule), amounts in dispute and cost risks. Concrete flat prices are permissible insofar as fee law allows and they do not undercut the statutory fees in an impermissible way. Phrase it transparently: 'The statutory fees depend on the amount in dispute' is a clean, AI-suitable statement that misleads no one.

Build your content in a clear question-and-answer structure. Language models most like to extract answers from texts that pose a question and answer it concretely in two to three sentences. That coincides with the factuality requirement: short, correct, verifiable.

What professional conduct law really prohibits with GEO

So you know the guardrails, here are the clear no-go areas. Promotional touting, misleading statements, unfair comparisons with named colleagues, and approaches directed at a specific mandate remain prohibited. So if you build AI-optimized landing pages, they may not contain untenable promises of success, invented specializations or disparaging statements about other firms.

A modern pitfall: AI-generated content on your own website. If you have texts created by ChatGPT, check them legally before you publish them. Models occasionally invent sections of law, deadlines or jurisdictions. If you publish such errors, that can be misleading in the sense of professional and competition law and, in the worst case, trigger a liability question. Your professional duty of care does not end at the prompt.

Collecting contact data via AI chatbots on your site is also regulated. Data protection under the GDPR and confidentiality apply from the very first message. Clarify transparently that a chatbot does not replace legal advice and does not establish an attorney-client relationship.

A practical roadmap: implementing GEO safely under professional law

Start with an honest inventory of your fields of activity and write a factual explanatory text for each. Add structured data such as the LegalService schema from Schema.org, so that search engines and AI capture your firm data unambiguously. That is pure technology and completely harmless under professional law, but it massively improves how correctly models reproduce you.

Keep your Google Business Profile, your entries in lawyer search services and your bar association details consistent. AI systems reconcile sources. Contradictory information - specialist lawyer here, not there - weakens your signal and can, in extreme cases, feed the accusation of misleading advertising. Consistency is both a GEO lever and professional-law protection.

Document which statements you make where. If a colleague or the bar association objects to a phrasing, you can prove that your source content was factual. You are not responsible for the AI output itself, but you very much are for your published source texts.

Conclusion: factuality is both professional conduct law and the best GEO strategy

The good news for you as a lawyer: what professional conduct law demands is almost identical to what makes good GEO. Factual, precise, verifiable, honest - by these criteria both Section 43b BRAO and modern language models assess your content. Whoever exaggerates promotionally loses twice: under professional law and in AI visibility, because models sort out unsubstantiated touting.

Don't see GEO as a risky gray zone, but as the consistent continuation of permissible firm advertising into the AI age. Your task is to describe your competence so clearly and correctly that ChatGPT, Perplexity and Google AI Overviews treat you as a reliable source. Stay the factual lawyer you are - then visibility is the logical consequence, not the risk.

If you are unsure, get an assessment from your bar association when in doubt, especially with new formats like chatbots or AI-assisted advisory previews. Professional conduct law keeps evolving, but the principle stays stable: factual information yes, loud touting no.

Focus-area statements in AI answers: where the line to the specialist title lies

When an AI answer names your firm, an area of activity often appears: "specialized in tenancy law" or "expert in traffic accidents". Here you have to be careful. Professional conduct law does allow you to name areas of interest and activity, but it demands a clear distinction from the true specialist-lawyer designation. Only someone formally entitled to hold the title should have "specialist lawyer for" (Fachanwalt für) in their content. Everything else belongs marked as a focus area.

In practice that means for you: consistently phrase it on your website as "area of activity family law" (Tätigkeitsschwerpunkt Familienrecht) instead of implying a specialist qualification that doesn't exist. AI systems often adopt your wording verbatim. If your source pages separate cleanly, the machine separates cleanly too. So check your legal notice, profile pages and structured data for whether specialist-lawyer titles without basis have crept in there. A single unclean building block can run through dozens of AI answers and earn you a cease-and-desist from colleagues.

Success figures and rates: why "95 percent won" is a risk

Clients like to ask AI systems for the "best" or "most successful" law firm. It is tempting to enrich your own content with success rates so the machine presents you as the winner. But blanket success figures like "95 percent of all cases won" are delicate under professional conduct law. They are hardly objectively provable, depend heavily on case selection and can quickly be rated as misleading advertising. What the AI makes of that, you cannot control.

Instead, rely on verifiable, factual information. Concrete, anonymized case examples with the actually achieved result are permissible and still give the AI usable context. Phrase it "In an unfair dismissal case we achieved a severance payment amounting to two gross monthly salaries" instead of a naked percentage. That keeps you within the factual frame while giving AI systems the structured, robust material they need for differentiated answers. Factuality beats superlatives technically as well.

Liability for false AI statements about your firm

A common worry is: what happens if an AI system claims nonsense about my firm, for instance a false focus area or a title I don't hold? In principle you are not liable for the independent errors of an external system. It only becomes critical if the false statement demonstrably rests on your own unclean content, or if you know of a possibility to correct it and remain inactive.

That's why a regular self-test belongs in your GEO roadmap. Ask the common AI systems specifically about your firm once a quarter and read the answers critically. If you discover a statement that is problematic under professional law, document it with date and screenshot and correct the underlying source. For persistent errors, use the providers' feedback functions. This clean documentation protects you twice: under professional law as proof of your diligence, and practically as a basis for improving your online presentation step by step.