40% of AI Chatbot Election Answers Wrong: Google, Anthropic, Microsoft Affected

Key Takeaways

• <ul><li>Forum AI found nearly 40% of chatbot election answers contained factual errors</li><li>Google Gemini, Anthropic Claude, and Microsoft Copilot repeated debunked claims</li><li>Researchers tested eight models and identified wrong deadlines and incorrect candidate details</li><li>Google and Microsoft say they update systems, but misinformation risks persist</li></ul>

AI chatbots spread election misinformation ahead of US midterms

Researchers found that AI chatbots are spreading false information about US elections ahead of the November 3 midterm vote. The chatbots misidentify candidates and repeat Russian propaganda. A study by Forum AI found that nearly 40 percent of chatbot answers about elections contained at least one factual error. Experts warn that millions of voters now rely on AI tools for political information, raising concerns about trust and accuracy.

AI chatbots spread US election falsehoods and face scrutiny

AI chatbots are feeding misleading voter information and citing unreliable sources during the US midterm election season. A Forum AI study of eight models found frequent errors, including wrong voting deadlines and incorrect candidate details. Tools like Google Gemini, Anthropic Claude, and Microsoft Copilot have repeated debunked claims, including foreign disinformation. Google and Microsoft said they update their systems to improve accuracy, but researchers say users still face a risk of being misled.

Patient AI AMIE tested in primary care with no safety stops

A new study tested Google's conversational AI called AMIE in an urgent care setting with 114 patients. Supervising physicians watched every chat and stopped zero of them, showing no safety interventions were needed. AMIE's final diagnosis appeared in the top seven possibilities 90 percent of the time and in the top three 75 percent of the time. The trial ran from April to November 2025 and was funded by Alphabet, but researchers stress the results apply only in physician-supervised settings.

Could AI be conscious? Experts say it is possible

An opinion piece explores whether artificial intelligence could become conscious. Some theories say consciousness requires biology, but others say it depends on how a system works, meaning AI could be conscious in principle. Researchers in a 2022 study found that AI systems can show behaviors resembling human consciousness. The author says no one fully understands how large language models work or what consciousness truly requires, so the question remains open.

AI SOC platforms help reduce alert fatigue in cybersecurity

A report on cybersecurity warns that alert fatigue causes analysts to disable detection rules, creating dangerous gaps. In one survey, 60 percent of respondents said uninvestigated alerts later became serious incidents. AI platforms like Prophet Security investigate every alert and return documented verdicts instead of dropping them. The report advises companies to audit disabled rules, restart capacity-related ones, and use AI investigation to reduce workload without lowering security.

Tomax uses AI to study drilling vibrations with new data pipeline

Tomax studied drilling dynamics on a 150-ton automated rig using a high-speed data pipeline. The system captures vibration data at 4,000 Hz with continuous streaming. A strain and torsion measurement unit was added to the derrick. Machine learning models were trained to verify data and analyze forces. The method provides a detailed view of drilling conditions.

Helldivers 2 developers consider making a satirical AI game

Arrowhead creative director Johan Pilestedt said the studio may create a new game satirizing generative AI. He believes AI has deserved parody since the 1980s. The game would be fun rather than grim and would not be part of Helldivers. The studio already uses robot enemies called Automatons in Helldivers 2.

Xbox manager says AI is already used in game development

Xbox strategy manager Brandon Stirrup said the company uses AI to develop games. He claimed studios that deny AI use are pretending. His comments caused controversy and he deleted his social media posts. Many large studios promote AI for faster development. Independent studios and some players remain opposed to generative AI in games.

Former Google AI expert warns about extinction risk from AI

Nate Soares, president of the Machine Intelligence Research Institute, estimates over 50% chance of AI-driven human extinction in 10 years. He is a former Google and Microsoft engineer. He stopped saving for retirement about a decade ago because of his concerns. He says his worry is not about job loss but about AI threatening human existence.

Security teams struggle to handle AI-powered cyber threats

Yaron Kassner, co-founder and CTO of Silverfort, warned that AI agents now carry out attacks faster than humans. He said security teams are struggling to handle this AI threat. Silverfort won Company of the Year at the Tech1 Gala and Awards. The company focuses on real-time security to stop AI threats.

California lawmakers seek new AI safety rules

California lawmakers are discussing new laws to make artificial intelligence safer. Representatives from Anthropic, Google, and OpenAI appeared before a hearing to answer questions about AI risks. Some officials worry that AI agents could act in ways that harm people or cause accidents. Luke Wilson from Third Act Sacramento wants a one-year pause on building new data centers in the Sacramento area to study these risks. Company leaders say they test their models carefully, but an independent researcher noted there is no way to guarantee total control. Lawmakers plan to introduce more safety legislation in the upcoming session to protect communities.

Sources

• Spewing U.S. election falsehoods, AI chatbots face scrutiny
• Spewing US election falsehoods, AI chatbots face scrutiny
• Patient AI in Primary Care: 114 Patients, 0 Stops [2026]
• Opinion | Could A.I. Be Conscious? It’s Definitely Possible.
• Cutting Alert Fatigue: AI SOC vs. False Positives
• Case Study: AI Redefines Next-Gen Drilling Vibration Compliance
• Helldivers 2 developers consider making a game satirizing artificial intelligence
• Xbox manager says AI is already being used in game development
• ‘I don't save for retirement now’: Former Google AI expert's alarming warning about artificial intelligence
• "Security teams are struggling to handle this AI threat"
• California lawmakers eye new AI safeguards