[{"data":1,"prerenderedAt":470},["ShallowReactive",2],{"insight-ai-invisible-participant-negotiations-fi":3},{"id":4,"title":5,"audience":6,"author":7,"body":8,"cover":456,"coverAlt":457,"data":6,"dateModified":6,"datePublished":458,"description":459,"extension":460,"locale":461,"meta":462,"metaTitle":463,"navigation":464,"path":465,"readingMinutes":466,"seo":467,"stem":468,"summary":20,"__hash__":469},"insights\u002Ffi\u002Fen\u002Finsights\u002Fai-invisible-participant-negotiations.md","AI: the invisible participant in negotiations",null,"Daryna Karuna",{"type":9,"value":10,"toc":438},"minimark",[11,15,18,21,26,29,32,35,102,111,114,118,121,124,127,198,201,204,207,213,217,220,223,227,230,233,237,240,243,280,284,292,296,301,309,313,316,320,323,327,330,334],[12,13,14],"p",{},"Picture a negotiation. Across the table sits an investor, a prospective partner or the buyer's lawyer. In front of them is a folder. It holds what they managed to find out about you in advance: company records, press coverage, what colleagues have said. And lately, one more page that nobody printed. It is an AI system's answer to \"what is known about this company?\" or \"who is this person?\". You have not seen that page, you do not know what it says, and most likely nobody will tell you about it.",[12,16,17],{},"Sometimes it is worse: the negotiation never happens. Someone read that page in advance, and you did not get the call. There was no meeting and no table, and you never saw the folder. You do not know what was in it, and you do not know why the talks fell through before they began.",[12,19,20],{},"Before a negotiation, the other side often asks AI about the person or company they are about to meet. What AI says influences the decision on a deal and can be the reason talks collapse or never start, and you will usually not find out. That risk can be measured and reduced before the counterparty sees it.",[22,23,25],"h2",{"id":24},"does-the-other-side-check-you-with-ai-before-a-negotiation","Does the other side check you with AI before a negotiation?",[12,27,28],{},"Before people negotiate with anyone at all, they ask AI about the person or the company. That is how negotiations acquired an invisible participant: it speaks about you before you have a chance to introduce yourself, and you hear none of it. I see this in how counterparty checks have changed, and the research points the same way. I believe almost every side in a negotiation now has a dossier like this. What AI says about a person or a company influences the decision on a deal and can be the reason it falls apart. That makes it a question of the company's economics.",[12,30,31],{},"In a G2 survey of 1,076 B2B software buyers conducted in March 2026, 51% said they now start their research with an AI chatbot more often than with Google. In G2's 2025 report, that figure was 29%. 71% of respondents use AI chatbots at some point in their research. According to TrustRadius (January 2026, 1,862 technology buyers), 63% used AI during their purchase journey.",[12,33,34],{},"Investors open the folder too. In late 2024 FTI Consulting surveyed 551 institutional investors based in the US, Europe and the Middle East: 93% of them use AI to analyse companies' reputations. The picture is similar for lawyers. In Thomson Reuters' 2026 report (a survey of 1,514 professionals in 27 countries, October to November 2025), 80% of legal professionals who use generative AI apply it to legal research, and more than half of all professionals surveyed use publicly available tools such as ChatGPT.",[36,37,38,54],"table",{},[39,40,41],"thead",{},[42,43,44,48,51],"tr",{},[45,46,47],"th",{},"Who opens the folder",[45,49,50],{},"Source and sample",[45,52,53],{},"Finding",[55,56,57,69,80,91],"tbody",{},[42,58,59,63,66],{},[60,61,62],"td",{},"B2B software buyers",[60,64,65],{},"G2, March 2026, 1,076 buyers",[60,67,68],{},"51% start research with an AI chatbot more often than with Google (29% in G2's 2025 report); 71% use AI chatbots at some point",[42,70,71,74,77],{},[60,72,73],{},"Technology buyers",[60,75,76],{},"TrustRadius, January 2026, 1,862 buyers",[60,78,79],{},"63% used AI during their purchase journey",[42,81,82,85,88],{},[60,83,84],{},"Institutional investors",[60,86,87],{},"FTI Consulting, November to December 2024, 551 investors in the US, Europe and the Middle East",[60,89,90],{},"93% use AI to analyse companies' reputations",[42,92,93,96,99],{},[60,94,95],{},"Legal professionals",[60,97,98],{},"Thomson Reuters, October to November 2025, 1,514 professionals in 27 countries",[60,100,101],{},"80% of legal professionals who use generative AI apply it to legal research",[12,103,104],{},[105,106],"img",{"alt":107,"height":108,"src":109,"width":110},"Poster: four hands with different sleeves pass a single terracotta card from one to the next. Headline: \"The file travels before you do\". Subline: \"One question to AI, and your file changes hands.\" Labels: \"investor · partner · lawyer · buyer\".",941,"\u002Fmedia\u002Finsights\u002Fneg-file-travels.jpg",1672,[12,112,113],{},"These figures come from technology buyers, investors and lawyers because those are the groups people measure most often. But I am convinced this applies to absolutely everyone: any company and any public figure. Anyone can ask AI about you before a meeting, and they do not need a procurement department to do it. The folder also has more than one reader. According to Forrester (January 2026), a typical B2B buying decision involves 13 people inside the company and 9 outside it.",[22,115,117],{"id":116},"what-can-ai-put-in-your-dossier","What can AI put in your dossier?",[12,119,120],{},"AI can put two kinds of error in your dossier: it confuses you with someone else, and it attributes someone else's facts to you. That is where I see the main errors. Two German court cases from 2026 show both.",[12,122,123],{},"In September 2026 the Regional Court of Aachen heard the case of a medical professor. Google Search's AI Mode was stating that he had left his clinic over the sexual coercion of his secretary and that a court had given him a suspended sentence. The court accepted the professor's sworn statement that all of this concerned a different former university professor. The court also knew that the criminal case had been brought against someone else. The professor learned about the text on 28 August from a fellow doctor and asked Google to remove it the same day. On 15 September Google replied that it had decided to take no action; that same day, the professor's lawyer sent Google a formal warning. On 24 September the court issued a temporary injunction without first hearing Google.",[12,125,126],{},"In May 2026 the Regional Court of Munich I dealt with a similar story involving companies. Google Search's AI Overviews described two publishing houses as \"known for dubious business practices\" (\"bekannt für unseriöse Geschäftspraktiken\"), claiming they lured customers into subscription traps and frequently changed their names and website addresses. The overview also linked them to other firms, and the court found no basis for that link. One of the sources the overview cited was a warning about someone else entirely. The court held that the overview was Google's own content, for which the company is responsible. The case later ended in a settlement.",[36,128,129,141],{},[39,130,131],{},[42,132,133,135,138],{},[45,134],{},[45,136,137],{},"Aachen",[45,139,140],{},"Munich",[55,142,143,154,165,176,187],{},[42,144,145,148,151],{},[60,146,147],{},"Court and date",[60,149,150],{},"Regional Court of Aachen, 24 September 2026",[60,152,153],{},"Regional Court of Munich I, May 2026",[42,155,156,159,162],{},[60,157,158],{},"Google feature",[60,160,161],{},"AI Mode",[60,163,164],{},"AI Overviews",[42,166,167,170,173],{},[60,168,169],{},"Who was affected",[60,171,172],{},"A medical professor",[60,174,175],{},"Two publishing houses",[42,177,178,181,184],{},[60,179,180],{},"What went wrong",[60,182,183],{},"Facts about a different former university professor were attributed to him",[60,185,186],{},"The companies were linked to other firms without basis; one cited source was about someone else",[42,188,189,192,195],{},[60,190,191],{},"Outcome",[60,193,194],{},"Temporary injunction without first hearing Google",[60,196,197],{},"Overview held to be Google's own content; the case later ended in a settlement",[12,199,200],{},"These errors have a mechanism. Cheng-Han Chiang and Hung-yi Lee of National Taiwan University showed that when names are ambiguous, language models merge facts about different people into one plausible biography, and standard factuality metrics miss it. The paper appeared in Findings of ACL 2024.",[12,202,203],{},"Besides confusion, the folder holds old material. In May to June 2025 FTI Consulting examined what ChatGPT said about DAX40 companies: 83% had at least one reputational risk, and outdated or long-refuted negative stories kept surfacing. In FTI's study of FTSE 100 companies (January 2026), 60% of the risks identified were legacy issues, ones the companies had often already addressed.",[12,205,206],{},"The danger lies in how people read these answers. In its 2024 report, the European Data Protection Board (EDPB) wrote that \"the outputs provided by ChatGPT are likely to be taken as factually accurate by end users, including information relating to individuals, regardless of their actual accuracy.\"",[12,208,209],{},[105,210],{"alt":211,"height":108,"src":212,"width":110},"Poster: a terracotta rotary telephone sits silent on an empty desk with the chair pushed in. Headline: \"The call that never came\". Subline: \"No meeting, no refusal, no reason given.\"","\u002Fmedia\u002Finsights\u002Fneg-call-never-came.jpg",[22,214,216],{"id":215},"why-will-you-not-find-out-what-the-other-side-read","Why will you not find out what the other side read?",[12,218,219],{},"The other side will not tell you what they read about you in AI. If the talks started, you will not learn why they stalled. If they never started, you will not even learn that they could have happened. Nobody will call, nobody will write, and nobody will sit down at the table with you. You cannot challenge a rejection that never took place. None of this will reach you. The professor in Aachen found out about the false claims only because a colleague told him.",[12,221,222],{},"You could object that buyers do not trust AI blindly. Forrester reports that buyers check AI answers with people they trust, and according to TrustRadius, 94% of those who used AI fact-check its responses at least some of the time. That is true. But in my view, people double-check what raises doubt. A confusion that looks plausible raises none. It will probably go unchecked, and nobody will ask you about it.",[22,224,226],{"id":225},"share-of-mentions-is-the-wrong-measure-for-negotiations","Share of mentions is the wrong measure for negotiations",[12,228,229],{},"For negotiations, the quality of what AI says about you matters more than the share of mentions. There is a lot of talk now about how the main thing is simply to get into AI. Success is measured in percentages: how often a company is mentioned, whether it made the list of recommendations. Tim Sanders, Chief Innovation Officer at G2, puts it this way: \"Winning the answer means being one of the recommended brands at the end of the report.\"",[12,231,232],{},"I disagree. If AI names your company often but confuses it with another or retells someone else's story, a high percentage works against you. Each of those mentions puts another wrong page in the folder.",[22,234,236],{"id":235},"why-checking-ai-answers-is-work-for-specialists","Why checking AI answers is work for specialists",[12,238,239],{},"Checking AI answers is work for specialists because a company or a person usually has no tool to measure those answers and find out why they look the way they do. You can ask ChatGPT a couple of questions about yourself. But one answer explains nothing: a different model, a different language or a differently worded question will give a different picture. Above all, the answer shows the problem but not its cause.",[12,241,242],{},"At QuorumLab we work from the premise that an AI answer is a symptom, and the cause lies in the information environment. We measure what is in this dossier, find the causes of the errors and help correct the information before negotiations. We do not control external AI systems, and we do not promise what words they will use about you.",[36,244,245,255],{},[39,246,247],{},[42,248,249,252],{},[45,250,251],{},"What AI does in your dossier",[45,253,254],{},"QuorumLab service",[55,256,257,269],{},[42,258,259,262],{},[60,260,261],{},"Confuses you with someone or links you to another company",[60,263,264],{},[265,266,268],"a",{"href":267},"\u002Fservices#companies-brands","Companies, Brands & Connections",[42,270,271,274],{},[60,272,273],{},"Repeats disputed or someone else's allegations as established fact",[60,275,276],{},[265,277,279],{"href":278},"\u002Freputational-forensics","Reputational Forensics",[22,281,283],{"id":282},"what-to-know-before-the-meeting","What to know before the meeting",[12,285,286,287,291],{},"The invisible participant reads about you before anyone schedules a meeting with you, and the folder on the other side is already assembled. A wrong page in this dossier is a risk that the negotiation will collapse or never start, and so a risk to the deal. You can measure and reduce that risk in advance, before the counterparty sees it. To do that, you need to know what is in the folder before the other side opens it: QuorumLab's ",[265,288,290],{"href":289},"\u002Fstart-your-check","free initial diagnostic"," is one place to start.",[22,293,295],{"id":294},"frequently-asked-questions","Frequently asked questions",[297,298,300],"h3",{"id":299},"can-i-check-what-ai-says-about-me-by-asking-chatgpt-myself","Can I check what AI says about me by asking ChatGPT myself?",[12,302,303,304,308],{},"You can, but a single answer tells you little. Another model, another language or a differently worded question can produce a different picture of the same company or person. A single answer also shows a problem without showing where it comes from. Finding the cause means measuring answers across systems and tracing the information behind them, which is the work described on QuorumLab's ",[265,305,307],{"href":306},"\u002Fservices\u002Fai-visibility-baseline","AI Visibility Baseline"," page.",[297,310,312],{"id":311},"is-google-responsible-for-what-its-ai-says-about-a-company-or-a-person","Is Google responsible for what its AI says about a company or a person?",[12,314,315],{},"Two German first-instance courts treated it that way in 2026. The Regional Court of Munich I held in May 2026 that an AI Overview was Google's own content; the case then ended in a settlement, so the judgment did not become binding. The Regional Court of Aachen issued a temporary injunction over AI Mode in September 2026. Neither decision is a final ruling on the merits.",[297,317,319],{"id":318},"why-does-ai-confuse-people-who-have-similar-names","Why does AI confuse people who have similar names?",[12,321,322],{},"When several people share a name, language models can merge facts about them into a single biography that reads as plausible. Cheng-Han Chiang and Hung-yi Lee of National Taiwan University documented this in Findings of ACL 2024 and showed that standard factuality metrics do not catch it. In the Aachen case, AI Mode attributed another professor's history to the claimant.",[297,324,326],{"id":325},"can-quorumlab-guarantee-what-ai-will-say-about-me","Can QuorumLab guarantee what AI will say about me?",[12,328,329],{},"No. QuorumLab does not control external AI systems and does not promise specific wording in ChatGPT, Gemini, Claude or Perplexity. The work is to measure what AI systems say, find what in the information environment shapes those answers, help correct the information where there is a basis to do so, and measure again to see what changed.",[22,331,333],{"id":332},"sources","Sources",[335,336,337,348,357,366,375,383,392,401,410,419,428],"ul",{},[338,339,340,341,347],"li",{},"G2: ",[265,342,346],{"href":343,"rel":344},"https:\u002F\u002Fcompany.g2.com\u002Fnews\u002Fg2-research-the-answer-economy",[345],"nofollow","The Answer Economy: How AI Search Is Rewiring B2B Software Buying"," (15 April 2026)",[338,349,350,351,356],{},"HG Insights \u002F TrustRadius: ",[265,352,355],{"href":353,"rel":354},"https:\u002F\u002Fhginsights.com\u002Fnews\u002Ftrustradius-2026-b2b-buying-disconnect-report-reveals-ai-has-changed-how-buyers-research-but-not-what-they-trust\u002F",[345],"2026 B2B Buying Disconnect Report"," (press release, 15 July 2026)",[338,358,359,360,365],{},"Forrester: ",[265,361,364],{"href":362,"rel":363},"https:\u002F\u002Fwww.forrester.com\u002Fpress-newsroom\u002Fforrester-2026-the-state-of-business-buying\u002F",[345],"The State of Business Buying, 2026"," (press release, 21 January 2026)",[338,367,368,369,374],{},"FTI Consulting: ",[265,370,373],{"href":371,"rel":372},"https:\u002F\u002Ffticommunications.com\u002Fwp-content\u002Fuploads\u002F2025\u002F07\u002FDAX40_LLM_Thoughtleadership_Whitepaper_final_DE.pdf",[345],"KI-Reputationsmonitor 2025: Was sagt ChatGPT über die DAX40-Unternehmen?"," (July 2025)",[338,376,368,377,382],{},[265,378,381],{"href":379,"rel":380},"https:\u002F\u002Ffticommunications.com\u002Fwp-content\u002Fuploads\u002F2026\u002F01\u002FFTI-Consulting_LLMs-and-FTSE-100_Jan-2026_brochure.pdf",[345],"AI Search revisited: Reputational risk in a fractured landscape"," (FTSE 100, January 2026)",[338,384,385,386,391],{},"Thomson Reuters: ",[265,387,390],{"href":388,"rel":389},"https:\u002F\u002Fwww.thomsonreuters.com\u002Fcontent\u002Fdam\u002Fewp-m\u002Fdocuments\u002Fthomsonreuters\u002Fen\u002Fpdf\u002Freports\u002F2026-ai-in-professional-services-report.pdf",[345],"2026 AI in Professional Services Report"," (2026)",[338,393,394,395,400],{},"Landgericht Aachen: ",[265,396,399],{"href":397,"rel":398},"https:\u002F\u002Fwww.itm.nrw\u002Fwp-content\u002Fuploads\u002F2026\u002F10\u002FKM_C450i26100808190.pdf",[345],"Beschluss, 8 O 268\u002F26"," (24 September 2026)",[338,402,403,404,409],{},"Landgericht München I: ",[265,405,408],{"href":406,"rel":407},"https:\u002F\u002Fwww.gesetze-bayern.de\u002FContent\u002FDocument\u002FY-300-Z-BECKRS-B-2026-N-11860",[345],"Endurteil, 26 O 869\u002F26"," (28 May 2026)",[338,411,412,413,418],{},"Chiang, C.-H. and Lee, H.: ",[265,414,417],{"href":415,"rel":416},"https:\u002F\u002Farxiv.org\u002Fabs\u002F2402.05629",[345],"Merging Facts, Crafting Fallacies: Evaluating the Contradictory Nature of Aggregated Factual Claims in Long-Form Generations"," (Findings of ACL 2024, pp. 2734–2751)",[338,420,421,422,427],{},"European Data Protection Board: ",[265,423,426],{"href":424,"rel":425},"https:\u002F\u002Fwww.edpb.europa.eu\u002Fsystem\u002Ffiles\u002F2024-05\u002Fedpb_20240523_report_chatgpt_taskforce_en.pdf",[345],"Report of the work undertaken by the ChatGPT Taskforce"," (para. 30, 23 May 2024)",[338,429,430,431,435,436],{},"QuorumLab: ",[265,432,434],{"href":433},"\u002Fservices","Services","; ",[265,437,279],{"href":278},{"title":439,"searchDepth":440,"depth":441,"links":442},"",2,3,[443,444,445,446,447,448,449,455],{"id":24,"depth":440,"text":25},{"id":116,"depth":440,"text":117},{"id":215,"depth":440,"text":216},{"id":225,"depth":440,"text":226},{"id":235,"depth":440,"text":236},{"id":282,"depth":440,"text":283},{"id":294,"depth":440,"text":295,"children":450},[451,452,453,454],{"id":299,"depth":441,"text":300},{"id":311,"depth":441,"text":312},{"id":318,"depth":441,"text":319},{"id":325,"depth":441,"text":326},{"id":332,"depth":440,"text":333},"\u002Fmedia\u002Finsights\u002Fneg-cover.jpg","An empty terracotta chair stands at the far end of a long negotiation table on a deep teal background. Two dark silhouettes sit on either side, and a closed folder lies on the table in front of the empty chair: the invisible participant in a negotiation is present but unseen.","2026-10-24","Before a negotiation, the other side often asks AI about you. What AI can put in that dossier, why you will not find out, and how to measure the risk.","md","en",{},"AI: The Invisible Participant in Negotiations",true,"\u002Ffi\u002Fen\u002Finsights\u002Fai-invisible-participant-negotiations",10,{"title":5,"description":459},"fi\u002Fen\u002Finsights\u002Fai-invisible-participant-negotiations","xZiGoZwqoH5n6Ka4WxuHqhNzxobIngrTdIIIVLSaNcU",1792855255671]