What to do when ChatGPT or another AI system gives incorrect information about you

Why this issue matters

An incorrect answer may arise from name confusion, conflicting sources, old data or an unsupported statement generated at that moment. It should be reproducible and documented.

The practical risk is not simply that a negative item exists. It is that a customer, employer, investor, patient, journalist or partner encounters it without enough context and uses it to make a decision. The relevant audience, query, market and source therefore matter as much as the content itself.

A responsible review starts by preserving the exact page, account, answer or search state. It then distinguishes verified fact, allegation, opinion, policy violation, personal data, identity confusion and an information gap. Those categories lead to different remedies and different limits.

Evidence to collect before acting

Record exact URLs, dates, screenshots, profile handles, search phrases and the country and language in which the issue appears. Capture the complete context rather than one cropped sentence. Where a statement is inaccurate, identify reliable evidence and the precise correction requested.

Do not send passwords, financial credentials, complete identity documents or illegal material to an adviser through a first-contact form. Sensitive evidence should be minimized and shared only through an agreed secure process if it becomes necessary.

• Exact public location of each item
• Date, language, market and affected query
• Source owner or platform
• Factual evidence and unresolved uncertainty
• Immediate safety or escalation needs

Choose the route that matches the problem

Save the exact prompt, date, language, model, answer and citations; correct controllable sources; use provider feedback or privacy processes where applicable.

Correction, removal, delisting, response, suppression and monitoring are not interchangeable. Removal changes the source; delisting may limit discovery for particular searches; response adds context; suppression develops stronger legitimate information; monitoring detects change. More than one route may be appropriate.

Platform policy and legal rights should not be used as generic pressure. A strong request identifies the specific rule or right, explains how the content meets it and provides enough evidence for an independent reviewer. Lawful criticism may remain even when it feels unfair.

Limits, measurement and the next decision

Publishers, platforms, search engines, regulators and AI providers make independent decisions. No responsible agency guarantees an external removal, a ranking, publication on Wikipedia or an exact AI response. Timing depends on the evidence, process and authority of the source.

Measure from a dated baseline. Useful indicators include correction or removal decisions, review status, source visibility, search composition, profile accuracy, response completion and repeated themes in AI answers. Separate completed activity from observable outcome and avoid claiming that one action caused every movement.

The next decision should be written down: what is being attempted, who approves it, what could amplify the issue, when to review progress and what alternative applies if the first route is rejected.

How to turn this guidance into a responsible plan

Begin with evidence, not assumptions. Save the exact URLs, screenshots, publication dates, search phrases, review profiles and AI answers that are creating concern. Record where each item appears, who controls the source and whether the information is inaccurate, outdated, private, misleading or simply unfavorable. These distinctions matter because removal, correction, response, suppression and monitoring are different remedies. A credible adviser should explain those differences before recommending work or discussing timing.

Next, define the audience and the decision at risk. A result seen by prospective clients in Spain may require different language, sources and local signals from a result affecting investors in the United States. Decide which names, brands, locations and search questions matter most. Prioritization prevents a campaign from becoming a vague attempt to control the internet and turns it into a measurable program focused on accuracy, trust and discoverability.

Evidence, people and measurements to prepare

Create a baseline before changes begin. It can include the first two pages of Google for agreed searches, ratings and review volume, visibility of owned pages, recurring themes in AI answers and the status of platform or publisher requests. Keep personal data to the minimum necessary and share sensitive documents only through an agreed secure process. If a legal right may apply, involve qualified counsel in the relevant jurisdiction; reputation strategy does not replace legal advice.

Assign an owner for approvals, factual verification and customer responses. Review progress consistently, but do not judge the program by one daily ranking. Useful measures include corrected or removed items, response completion, the share of credible owned and independent sources, search-result composition, branded query trends and whether public information answers real questions. The objective is a more accurate and resilient digital record, not an artificial promise that criticism will disappear.

Keep a decision record, not just a list of links

For every material item, record the exact source, the factual concern, the person responsible for verification, the proposed route and the reason that route is proportionate. Include the date of any request, response deadline, platform reference number and next review date. This record prevents duplicate or contradictory reports and helps a new decision-maker understand why an item was corrected, challenged, answered, monitored or left alone.

A sound record also separates confirmed facts from interpretations. Label legal questions for qualified counsel, service failures for operational owners and public explanations for communications review. If an item changes, preserve the before-and-after evidence without collecting more personal data than necessary. This discipline is especially important when search engines and AI systems update unpredictably: it shows what actually changed and avoids claiming credit for movements that may have several causes.

• Source, URL, date and affected search or prompt
• Verified facts, disputed claims and supporting evidence
• Chosen route, owner, status and next review date
• Privacy, amplification and escalation risks