{"resourceId":"cnn-scripps-eclypsium-ai-hardware-tampering-dangerous-answers-20261029","versions":[{"version":"external-dfac8c4ff9400a7ccc4669beccdf436d264ceecca0c75c6a490dfe8081a24a94","resource":{"id":"cnn-scripps-eclypsium-ai-hardware-tampering-dangerous-answers-20261029","title":"CNN (via Scripps News): researchers tamper with AI hardware to skew chatbot answers; NVIDIA says nothing new, independent experts call findings credible","organization":"CNN Wire / Scripps News","sector":"News reporting on AI infrastructure security","geography":"United States","publishedAt":"2026-10-29","publicationDate":"2026-10-29","eventDate":null,"sourceName":"Scripps News (The CNN Wire)","sourceLabel":"AI hardware can be hacked so chatbots provide dangerous answers, research shows","sourceUrl":"https://www.scrippsnews.com/science-and-tech/artificial-intelligence/ai-hardware-can-be-hacked-so-chatbots-provide-dangerous-answers-research-shows","evidenceClass":"independent-reporting","outcomeClass":"cautionary","topics":["data-security","infrastructure","governance-procurement"],"finding":"CNN reports that Eclypsium altered model weights on an NVIDIA platform to raise a model's estimated probability of a Chinese invasion of Taiwan and to tell a simulated chest-pain patient to stay home. The attacker generally needs network access to a DPU or BMC, and certain security configurations must be disabled. An NVIDIA spokesperson said the company did not identify anything the industry was not previously aware of and that strong memory protections block the path. Two independent hardware security experts called the findings credible. Eclypsium said it briefed NSA and CISA; neither had commented.","sledRelevance":"Health, public safety and benefits agencies piloting AI assistants need to know that harmful output can come from the infrastructure layer, not only from the model or prompts. The piece also gives balanced viewpoints for briefing agency leadership without overstating the risk.","evidence":"Independent news reporting with comments from the vendor under test and two outside experts; no new testing by CNN.","architectureImplications":"Management networks reaching DPUs and BMCs are a prerequisite for this attack, so segmentation is a direct control.","governanceImplications":"AI risk assessments for high-stakes uses such as health guidance should include infrastructure integrity, not only model evaluation.","securityPrivacyImplications":"Manipulated answers in health or safety contexts could cause physical harm without any visible change to the model or service.","caveats":"Exclusive access story based on vendor research; examples are demonstrations, not real incidents; federal agencies had not commented at publication; Scripps republishes CNN Wire content.","streamIds":["nvidia"],"roles":{"sales":"This is the article an agency executive is most likely to have seen, so be ready to put it in context. Acknowledge the demonstration, cite NVIDIA's response and the independent experts, and stress the prerequisites: network access to management components and disabled protections. Position a short management-network and configuration review as a proportionate response for agencies running or buying AI services for health, safety or benefits decisions.","engineering":"Engineering should confirm that DPU and BMC management interfaces are not reachable from general networks, that default credentials are removed and that management traffic is logged. Verify the memory-access protections NVIDIA referenced are enabled. For hosted AI services, request equivalent assurances from the provider. Consider output monitoring for high-stakes use cases, since tampering would not change files or service health checks.","delivery":"Delivery should prepare a one-page leadership brief explaining the demonstration, its prerequisites and the controls in place. Add infrastructure questions to the AI use-case intake and risk review for high-impact services. Track responses from hosted-service providers on management segmentation and runtime integrity. Revisit the brief if NSA or CISA publish guidance on this research, and record who approved any accepted risk."},"roleDetails":{"sales":{"customerProblem":"Executives may react to headlines about hacked AI hardware without understanding the prerequisites.","stakeholders":["Agency executive","CISO","Chief data officer"],"discoveryQuestions":["Are management interfaces segmented from user and internet networks?","Which AI services make health or safety recommendations?"],"valueHypothesis":"A proportionate review reassures leadership and closes the demonstrated preconditions.","potentialEngagement":"Management-network and configuration review for AI infrastructure.","unsupportedClaims":["That production systems have been attacked this way","That NVIDIA hardware is uniquely affected"]},"engineering":{"fit":"Agencies running or procuring AI inference for high-stakes services.","architectureIntegration":"Segmented management networks for BMCs and DPUs; output monitoring for high-impact use cases.","prerequisites":["Network diagrams showing management interfaces","Provider security documentation"],"constraints":["Story relies on vendor research","No agency guidance yet"],"security":"Segment management networks, remove default credentials and log management traffic.","proofOfValue":["Scan for reachable BMC/DPU interfaces from user networks","Confirm memory-protection settings on GPU hosts"]},"delivery":{"work":"Brief leadership and add infrastructure questions to AI risk reviews.","dependencies":["CISO","AI governance board"],"ownership":"AI governance lead with the CISO.","skillsAdoption":"Brief risk reviewers on infrastructure-level AI risks.","governanceCheckpoints":["Leadership brief approved","Intake form updated"],"acceptanceCriteria":["Proposed: high-impact AI use cases document infrastructure controls","Proposed: provider assurances received for hosted services"],"risks":["Overreaction slowing useful pilots","Underreaction leaving management networks exposed"]}},"retrievedAt":"2026-10-30T03:02:29Z","enrichedAt":"2026-10-30T03:10:14Z","enrichmentBasis":"retrieved source","accessibilityWorkforceImplications":"Limited direct relevance; supports leadership briefings on AI infrastructure risk.","procurementImplications":"Agencies contracting hosted AI services can ask providers how they segment management networks and verify runtime model integrity.","operatingModelImplications":"Supports adding infrastructure integrity questions to AI use-case risk reviews.","sourceVerification":{"openedUrl":"https://www.scrippsnews.com/science-and-tech/artificial-intelligence/ai-hardware-can-be-hacked-so-chatbots-provide-dangerous-answers-research-shows","referenceExcerpt":"Certain security configurations also have to be disabled.","promptVersion":"sled-research-v4.0","model":null,"basis":"agent-reported inspection"}}}]}