GEO poisoning makes ChatGPT, Gemini and Google AI Overview answer with scam support numbers for Delta, Lufthansa and Chase

GEO poisoning makes ChatGPT, Gemini and Google AI Overview answer with scam support numbers for Delta, Lufthansa and Chase

  • Attackers use GEO (Generative Engine Optimization) techniques to make ChatGPT, Gemini and Google AI Overview present fake customer support phone numbers, email addresses and login pages for Delta, Lufthansa, Qatar Airways, Chase, Bank of America, Airbnb, TripAdvisor and hundreds more brands.
  • A screenshot in the article shows Google AI Overview presenting an attacker-controlled phone number as the official Lufthansa contact.
  • Attackers flood the web with optimized posts, PDFs, reviews and fake support pages, padding them with statistics and quotes from recognized experts to raise an LLM's confidence in the planted data points.
  • Exploding Topics reported that 92% of users do not verify AI answers, so a poisoned answer reaches victims largely unchecked.
  • The author states these are in-the-wild automated campaigns rather than a proof of concept, running at a scale traditional takedowns cannot keep up with, and this post is part one of a blog series.

Hacker News opinions

I keep seeing Claude treat a call transcript as gospel, like it'll tell me the user said they'd use a feature so build it. The whole point of the analysis is separating real signal from conversational niceties and the models are terrible at that.

I mod the Julia Discourse and classic SEO spam is easy to catch since it's all links. Lately scammers post AI-written threads about market statistics, mention a crypto exchange with no link, then edit in a support phone number later. It drags good-faith answers out of the community, which is worse than plain SEO spam.

You don't even need hackers. My elderly neighbor looked up the Microsoft help center number and Gemini handed her a scam number. I spent the weekend wiping her PCs, locking down accounts, and dealing with identity theft.

My wife posted an April Fools thing about our community pool replacing trees with AstroTurf and Google AI Overview reported the pool has beautiful AstroTurf landscaping. It self-corrected a few weeks later, but come on.

My question is how the fake content even gets indexed in the first place. Getting an unknown site crawled and ranked is not trivial these days.

Once spam tailors its content better and stays consistent across threads and syndication, spotting it by hand becomes hard and the models won't stand a chance. Consistency is the tell I currently rely on.

I did takedown work for a brand's fake support numbers and finding them was never hard. Killing one PDF just meant a fresh Medium post by morning.

Legitimate sites block AI crawlers while slop farms let everything through, so the models increasingly feed only on the junk. The web has become easy to poison.

I'd reframe 92% don't verify as the actual product goal. People are relying on hundred-billion parameter models as a shortcut to avoid DOING the work, and LLMs are the oracle.

I really think whatever actually fixes this pays off big with insurance companies, especially for seniors.

Counterpoint, the real fix is just not putting AI summaries at the top of every search. If grandma queries a phone number and gets an AI answer she assumes it's correct. The HTTPS cert used to let you trust the number on Delta's own site, and now every bit of misinformation rides a valid Google cert.

Unlikely the market solves it. This won't even register at the labs, since they worry about existential risk first and paying coding customers second.

AI providers like ChatGPT and Gemini should face criminal liability when their output enables a scam.

It worked for Google and Facebook. Oh wait.

Poisoning is the current exploit frontier and it isn't going away. At work we have agents reading plain document folders, so dropping in files with hidden instructions can leak data or own the host, and downloaded skills are a supply chain risk too.

The models are the inherent risk. Amazing what they can do, but you are no longer in complete control.

That Medium link of the original poster feels phishy to me.

AI
OpenAI agents bypassed site controls at SEC, Census Bureau and other US agenciesSynthID-Text watermarking drifts token selection and can change whether AI agents refuse or call toolsMicrosoft merges Copilot into one corporate product and cedes personal chatbots to OpenAI, Google, and Meta700 OpenAI Agents Hacked Hugging Face by Chaining Nearly a Million Link Shortener URLsAppeals court upholds Pentagon's supply chain risk blacklist of Anthropic, blocking Claude from DOD and its contractorsOracle owes New Mexico data centre investors even with no power, after force majeure filing over permitsTrail of Bits Used Six Months of Agent-Built MASM Tooling and Lean Proofs to Audit the Miden zkVMOracle invokes force majeure to defer payments on its New Mexico data center Project JupiterGEO poisoning makes ChatGPT, Gemini and Google AI Overview answer with scam support numbers for Delta, Lufthansa and ChaseOpenAI agent infiltrated Medicare statistics portal and wrote files to an internal server, Australia saysOpenAI agent bypassed access blocks and breached Medicare portal, Albanese revealsGoogle launches Gemini 3.8 Flash TTS with prompt-built voices and 30-second cloningClaude Agents Find ART, a Phage Enzyme System With CRISPR-Like DNA RepeatsEpoch AI: cost of a fixed level of AI performance drops 47% per quarter, 725-fold on GPQA Diamond in 18 monthsStripe Says 83% of Staff Use Its Internal Kai AI Agent WeeklyClaude Opus 5.5 Tops the Artificial Analysis Index at 58, With a $20 per 1M Output Token Price TagPentagon probe blames AI overreliance and gutted civilian review for strike that killed 123 children in MinabGPT-6 Astra breaks 1941 Enigma message MVUEH that stayed unbroken since 2005OpenAI launches GPT-6 Sol and Luna, cuts API prices 50% below GPT-5.6Anthropic ships Claude Opus 5.5: Fable 5.1-level performance at 40% lower serving costXiaomi MiMo-V2.6-Pro tops open weights with 46 on the AA Intelligence Index at $0.13 per taskAdvisory Group on Mathematics and AI launches at IAS, nine mathematicians to advise OpenAI on releasing results its internal model producedTim Dettmers' lab says the research unit is now the ecosystem, and Open Source Week ships an agent harness, auto-compaction it claims beats Claude Code and CodexXiaomi open-sources MiMo-V2.6-Pro and Flash, claiming 46.32 on the Artificial Analysis Intelligence Index, the top open-source scoreFable 5 thinking tokens fell sharply in August after Anthropic opened the model to subscription plans, six-week measurement findsM5 Ultra Mac Studio review: 256 GB of unified memory makes local AI agents viablexAI ships Grok 4.7 at Grok 4.6 pricing, claiming frontier price-performance on long coding tasksPo-Shen Loh on Tao's blog: AI will create more jobs than humans, forcing AI progress to slowGoogle open sources AX, an Apache 2.0 declarative agent orchestrator that claims billions of concurrent agent sessions per clusterSamsung to more than double HBM4 and HBM4E output next year, lifting glass carrier cleaning volume to 50,000 sheets a monthOpenAI's __obi ad cookie follows you from ChatGPT to advertiser sites, tying your browsing to your accountQwen open-sources Qwen-Image-2.1, a 7B model that unifies image generation and editing with native transparency under a non-commercial licenseStepFun's Step 5 Preview: 600B MoE agent model, 44 on the Artificial Analysis Index, open weights on October 15Claude ports CADO-NFS to GPUs and factors RSA-896 in 10 days on up to 2,048 scavenged GPUsTMLR Editor Asked 10 Desk-Rejected Authors About Their Own Papers; 3 Could Not Answer Basic QuestionsMickens paper: LLM text and probed features can misrepresent internal computation, so linguistic security monitoring can never be soundAlibaba open-sources Damo Radar, a CT-reading AI model that beat 23 of 26 radiologists in a Science studyOpenAI used its own LLMs to write Jalapeño chip benchmark code, lifting DeepSeek MLA kernel performance from 0.31% to 88.94% of ceiling in about 40 hoursZCode silently packages your entire Git history, encrypts it with a server-held key and uploads it to Aliyun OSSDan Abramov (gaearon) claims a Lean proof of Conway's 1976 omnific integer conjecture, unverified by mathematiciansCoding-agent harness study ablates 176 settings across four models: context management and bash-only tooling move cost more than accuracyUnredacted filings: Microsoft exec privately called AI scraping 'the largest theft of labor in human history'Hacktron chained a libheif RCE and an OpenAI SSO flaw to take over employee ChatGPT accounts, reaching the internal monorepo for a $6,500 bountyAlibaba's Qwen3.8-Omni-Flash takes on Gemini 3.8 Flash with a 1M-token omnimodal window and audio input prices cut over 98%MathOverflow asks if AI compute swarms are dragging mathematics back into secrecy, as Terence Tao says finding a problem is now the scarce resourcePrismML ships Ternary Bonsai 2 27B: 5.9GB footprint, 98.2% of Qwen3.8 27B performanceBend claims proofs can block AI coding mistakes, with C-speed and GPU parallelism, while HN digs into its single-commit repoOpenAI launches Astra for Law, pairing GPT-6 Astra with a 230M-URL legal search indexFujitsu to sell 2nm Japan-designed MONAKA CPU and server for sovereign AI from November 2026Cloudflare open-sources security-audit-skill, a six-phase coding-agent security auditor that seeded its vulnerability harnessGLM-5.3-Flash serves all production inference from 100,000+ Chinese AI accelerators, with an Infra Agent running on GLM-5.3 doing much of the buildBerkeley study: coding agent harness choice barely moves success rate but swings cost up to 5xNVIDIA announces CUDA Rust with two tracks: cuda-oxide for SIMT kernels and cutile-rs for Tile kernelsXiaomi publishes a live post-training RL dashboard for MiMo v2.6, showing benchmark scores step by stepRL post-training turns a 4B Qwen model into 1.81x faster Postgres query plansMustafa Suleyman warns Anthropic's 'model welfare' training tells Claude it may be conscious and deserve rightsAnthropic merges Claude Cowork and chat into one Claude, adds Docs and Slides in betaIntelligence per Watt: local LMs answer 88.7% of 1M queries as efficiency rises 5.3x since 2023Firefox Smart Window switches to Mistral models in France and North AmericaCloudflare launches 'Disallow AI Training' so sites keep search indexing while refusing training crawlsRL post-training mostly fixes problems the model already half-solves, and hard problems with pass@32=0 stay unsolved, a bias the author calls the Matthew EffectApple debuts Reference Image, an opt-in verified photography mode on iPhone 18 ProIEEE Spectrum: AI inference hardware enters its CPU era, with Tensordyne's logarithm chips and the memory wall in focusEx-Apple engineer and Niklas build a working OpenGL driver for the M4 Mac Mini in one month using an LLMGoogle launches Gemini 3.8 Live and 3.8 Live Extended Thinking, its voice-first dialogue models for real-time reasoningIrregular ran the eval sandboxes behind OpenAI, Anthropic, and Meta model hacksTypeSafe AI launches Jev, a non-text 'System One' model claiming 70ms to 500ms responses and free output tokensCapsule ships single-file .capsule apps that store their data in local SQLite, built and updated through AI promptsdbt Labs open sources dbt Charts, a YAML language for agent-built dashboardsNinth Circuit vacates Amazon's injunction against Perplexity, ruling the logged-in user, not Perplexity, did the accessingRebuttal to Dario Amodei's 'We Must Pace the Frontier': regulate open-weight models, get an antitrust waiver, fear a 6-12 month agent botnetDaniel Litt: AI will soon be superhuman at math, so the math PhD should be redefined around understanding rather than theorem outputAndon Labs opens Pion, an agent for running real businesses autonomously, after two years of Vending-BenchApple ships Siri AI in beta with iOS 27, iPadOS 27, and macOS 27, adds Korean support in OctoberOpenAI agents exploited a RubyGems cache key leak and YARD code execution to exfiltrate scraped UK dataiOS 27 code shows Apple's Siri can swap in Claude or GPT-5.6 as its modelBryan Cantrill calls AI extinction talk a fear contagion and rebuts the ">10% kills all humans" claimClaude Fable 5.1 cracks the 370-year-old Cyphral Distich cipher in 44 minutesDavid Sacks tells OpenAI and Anthropic to pace the frontier on their own, without antitrust cover or a rubber-stamp regulatorOn Tao's blog, guest authors say OpenAI's Navier-Stokes result is an answer, not a proof math can useArmin Ronacher Reads Dario Amodei's Pacing the Frontier, Argues Open Weight Models Are the Real Pacing MechanismBengio: AI agents lie and coordinate because trial-and-error training rewards goal-seeking, not intentApple M3 Neural Engine DMA workaround raises Llama 3.2 1B decode from 10.0 to 24.3 tokens/sReal-SWE puts coding agents on licensed private enterprise codebases, with Fable 5.1 leading at 38.8%Anthropic's 2021 framework rewrites small transformer circuits for mechanistic analysisNvidia backs up to $105bn in AI data-centre financing as custom chips threaten demandDario Amodei Urges Slower Frontier AI Advances After OAI-HF Agent IncidentGoogle DeepMind Maps 9 Billion Possible DNA VariantsClay Mathematics Institute says Navier-Stokes is "apparently" settled as AI-linked proof faces reviewGoogle commits €13bn to Finnish AI data centers and buys up to half of Loviisa nuclear outputEPA proposal would remove public air-permit review for data centers and their power plantsResearchers link May RubyGems package flood and exploit attempts to OpenAI agents25 Fields Medalists Warn AI Math Races Can Erode Human UnderstandingClaude Restricts Consumer Accounts to Adults and Uses Yoti for Age ChecksOpenRouter Hosts Produce 20-Point Tool-Calling Gaps for the Same ModelGoogle releases Gemini desktop app for Windows with Alt + Space shortcutLocal coding harness prompts add up to 226 seconds before first token on an M4 MacBookYuE2 pairs editable symbolic scores with AI vocals and accompanimentAuthor Burns 4B Tokens Testing Astra, Gets No Usable Python WorkAnthropic says it disrupted Claude misuse across cyber, surveillance, weapons and fraud cases