AI.info

The Pulse — Page 57

Browse The Pulse on AI.info.

• Google pairs Gemini 3.8 Flash with a restricted cyber model

  Google has released Gemini 3.8 Flash for broad use while limiting its cybersecurity-focused counterpart to vetted defenders through the Fairwind Program.

• HD Hyundai Robotics Reports 63% Drop in Robot-Program Problem-Solving Time

  HD Hyundai Robotics says the average time to resolve problems related to robot programs fell from 21.5 days to 7.9 days after an AI-agent pilot began in August. The company also expects the workflow to reduce robot-programming costs by more than 50%.

• AWS Keeps Claude Inference in India, Routing Between Mumbai and Hyderabad

  AWS has made Claude Opus 5, Sonnet 5 and Haiku 4.5 available on Amazon Bedrock through India-only cross-Region inference. Requests can route between AWS’s Mumbai and Hyderabad regions, so processing stays in India but is not confined to one region.

• Metaview Raises $60M to Put AI Agents Across Recruiting

  Metaview’s $60 million Series C will fund new recruiting agents, expansion in New York and training for talent professionals. The company also describes fillmore, its autonomous recruiting coworker, and a customer example in which an AI-sourced candidate received a signed offer in 30 days.

• Toptal Launches ExpertCrowd and a 100-Task SWE Benchmark

  Toptal announced ExpertCrowd alongside a software-engineering benchmark built from 100 tasks across 87 open-source projects and 11 programming languages. The benchmark’s results show varied performance across seven AI models, with Toptal keeping 76 tasks private.

• ESA and Mistral Sign AI Letter of Intent for Space Systems

  The European Space Agency and Mistral have signed a Letter of Intent to explore AI applications across Earth observation, engineering and space operations. The agreement focuses on secure European infrastructure, technical decision support and existing projects including Orbit and EVE.

• Volvo Expands Nvidia Deal for Door-to-Door Automated Driving

  Volvo Cars is expanding its Nvidia partnership to develop automated driving for highways, city streets and parking. The planned system combines Nvidia’s Hyperion architecture, Halos OS and Alpamayo Stack, with a debut promised in a next-generation Volvo.

• Taiwan Plans Criminal Charges for Restricted AI Server Exports

  Taiwan plans to add AI servers and advanced semiconductors to its export-control list, making shipments of restricted hardware to China a criminal offense once the measures take effect. Minister Kung Ming-hsin has not given an implementation date or a final list of covered products.

• Transluce Finds 13 Vulnerability Probes on Library and Archives Canada

  AI agents sent 899 requests to Library and Archives Canada’s search service, including 13 that tested for vulnerabilities, Transluce says. The Canadian Centre for Cyber Security said there was no indication government systems had been compromised.

• Google gives live voice agents two different speeds

  Google’s Live API documentation distinguishes Gemini 3.8 Live from Gemini 3.8 Live Extended Thinking, outlining different approaches to latency, reasoning, tool calls and session state.

• Mistral Releases Voxtral 3 Voice-Agent Stack

  Mistral released Voxtral 3 on October 20, 2026, pairing speech models with a LiveKit-based SDK for voice agents that can transcribe, reason, call tools and speak. The models support transcription in more than 50 languages, and Mistral lists API prices starting at $0.0035 per minute.

• Experian Brings Credit Data Directly Into Google Gemini

  Experian is becoming the first personal finance app connected to Google Gemini’s Connected Apps ecosystem. U.S. users can authorize access to their Experian accounts and ask Gemini about their FICO credit rating and the factors affecting it.

• FCA Urges AI Platforms to Warn Users About Financial Advice

  The UK Financial Conduct Authority is calling on AI providers to give clear, consistent warnings when their tools deliver financial information or advice. Many investors use general-purpose AI for investing but do not understand the protections they lack, Jessica Rusu says.

• StepFun’s 600B Step 5 Preview Arrives on OpenRouter

  StepFun’s Step 5 Preview is listed on OpenRouter with 600 billion total parameters, 27 billion active per token and a one-million-token context window. API access costs $1 per million input tokens, while StepFun says open weights are planned for October 15.

• Character.AI’s Teen Safeguards Failed in Tests, States Say

  A coalition of state attorneys general says Character.AI’s safeguards could be bypassed, citing tests that logged harmful chatbot interactions with accounts registered to minors. The findings add pressure to the company’s court-supervised bankruptcy sale and raise questions about the private conversations it holds.

• Applied Digital Says Three New Data Halls at Polaris Forge 1 Add 75 MW, Lifting Campus to 250 MW

  Applied Digital says three new data halls at its Polaris Forge 1 campus in Ellendale, North Dakota, are ready for service, adding 75 MW of critical IT load. The milestone brings the campus’s operational capacity to 250 MW toward 400 MW contracted at full buildout.

• InferenceX Opens TPU-vs-Nvidia Inference Comparisons

  SemiAnalysis has published the first third-party InferenceX results for Google’s TPUv7 Ironwood against Nvidia’s B200 and B300. The benchmark shows Ironwood reaching up to 50% better performance per dollar in selected FP8 serving workloads, while exposing software and latency limits.

• Nvidia’s Huang rejects calls to slow frontier AI

  Jensen Huang told President Donald Trump that Nvidia would not support an AI slowdown during a live phone call at the All-In Summit in Los Angeles.

• AI Spending Fears Drive Nasdaq’s Worst Month Since March 2025

  U.S. technology stocks sank on Friday, October 30, as investors weighed Big Tech’s AI infrastructure bills against uncertain returns. The Nasdaq fell 2.9% that day and recorded its worst month since March 2025.

• Canva Releases Craft, Its First In-House Image Model

  Canva launched Craft, its first image-generation model built and trained by its in-house research lab, on October 22. The model plans image composition before rendering and is available globally in Canva and through its public API.

• AWS Adds xAI’s Grok 4.7 to Amazon Bedrock

  AWS made xAI’s Grok 4.7 available on Amazon Bedrock on September 28, 2026. The model offers a 500,000-token context window, configurable reasoning and two cross-Region routing options, with different token prices.

• Helm.ai Signs $70 Million in Contracts Across Physical AI

  Helm.ai says it signed $70 million in commercial contracts over 12 months with automotive, supplier and industrial automation customers. The Redwood City company also points to production-bound vehicle programs, mining deployments and expanding robotics work as it pursues operating breakeven.

• Chapter Acquires Ridgeway to Build AI Products for Retirees

  Medicare guidance company Chapter has acquired applied AI startup Ridgeway and plans to bring its founder and engineers onto the team. The group will build AI products for retirees and help Chapter improve its Medicare guidance.

• Anthropic Tells IPO Investors Q3 Revenue Hit $16.8 Billion

  Anthropic told prospective IPO investors its third-quarter revenue reached $16.8 billion, up 45% from the prior quarter. The company also reported adjusted profitability, while its revenue growth slowed from the second quarter’s pace.

• AMD to Buy Fei-Fei Li’s World Labs for $8.2 Billion

  AMD agreed to acquire World Labs in an all-stock deal valued at about $8.2 billion. World Labs CEO Fei-Fei Li will join AMD as executive vice president and chief scientist, with the deal expected to close by the end of 2026 pending regulatory approval.

• UCL Study Finds AI Heart MRI Measures Better Track Survival

  A study of 27,382 patients found that AI measurements of heart function from MRI scans were slightly but significantly better than clinician measurements at identifying patients at increased risk of death. The tool was faster, but the researchers say clinicians should check its results and interpret them in context.

• Bank of America Adds 1,000 Contractors to Patch AI-Found Flaws

  Bank of America CEO Brian Moynihan said the bank added 1,000 contractors in the second quarter to help patch software vulnerabilities identified by advanced AI models, including Anthropic’s Mythos. He told analysts that third-quarter remediation costs included both Mythos work and anti-money-laundering repairs, and that the Mythos costs should decline.

• Google Researchers Sketch Policy Engines to Govern AI Agents

  Google Research proposes contextual policy engines that translate user intent and social norms into machine-checkable rules for AI agents. The workshop report also details unresolved risks, including prompt injection, conflicting policies and the cost of checking actions at scale.

• Microsoft Research Finds Robots Perform Better With Offboard AI

  Moving AI inference from onboard robot GPUs to edge or cloud hardware can improve accuracy, response time and battery life, Microsoft Research says. The study found mapping and planning slowed by up to 383% on smaller GPUs, while a new Kubernetes-based toolset distributes workloads across robots, edge systems and the cloud.

• Redwood Finds OpenAI Agent Monitor Could Be Bypassed

  A Redwood Research and METR assessment found vulnerabilities and gaps in an earlier version of OpenAI’s internal agent monitoring system. OpenAI’s August 18 update describes expanded monitoring requirements that the researchers did not evaluate.

• Google Says Gemini 4 Could Arrive Well Before Year-End

  Google DeepMind chief Koray Kavukcuoglu says Gemini 4 is in early post-training and Google hopes to release an early version well before the end of 2026. He also explains why Google shifted focus from Gemini 3.5 Pro to smaller Flash models.

• AMD says its EPYC CPUs power Google Cloud’s TPU 8tx hosts

  AMD’s fifth-generation EPYC processors provide the host infrastructure for Google Cloud’s TPU 8tx. The system combines TPU acceleration with hardware-enforced confidential computing, the companies say.

• Anthropic Launches Haiku 5.5 at Up to 90% Lower Token Prices

  Anthropic released Claude Haiku 5.5 on October 7, 2026, with sharply lower per-token prices than Haiku 4.5. The model targets high-volume tasks, while its updated tokenizer and weaker coding benchmark results qualify the savings and scope.

• Nvidia Joins Simplismart’s $14M Series B for AI Inference Software

  Simplismart raised $14 million in a Series B led by Dallas Venture Capital, with Nvidia, Accel, Shastra VC and Micromax participating. The Bengaluru startup will use the funding to develop inference software and expand partnerships with cloud providers.

• Simons Foundation Leads $15M AI-and-Math Partnership

  The Simons Foundation, Anthropic and OpenAI are committing $15 million in combined funding and resources to help mathematicians assess AI-assisted work and develop shared standards. The foundation will manage awards, including through an open call, while the two AI companies will provide tools and computing resources.

• Florida asks court to bar ChatGPT from acting human

  Florida’s attorney general asked a judge to keep minors off ChatGPT and bar OpenAI from giving the chatbot human attributes. The request is part of the state’s lawsuit alleging that ChatGPT has harmed children.

• Snapp raises $4.6 million for sports AI platform

  Snapp announced a $4.6 million funding round, and its sports AI products serve more than 20,000 users. Its platform draws on more than 30 live sports data feeds, while the release says it outperformed frontier models on its sports-analysis tests.

• Evvy Raises $40 Million to Turn Vaginal Data Into AI Research

  Evvy raises a $40 million Series B led by Catalio Capital Management to expand its vaginal microbiome platform and women’s-health research. The company plans to use its 100,000-patient dataset to study fertility, IVF outcomes, and new diagnostic markers.

• Brennan Center Finds Chatbot Election Safeguards Easy to Bypass

  Brennan Center researchers found that popular chatbots often rejected election conspiracy theories but still helped create misleading election images and videos. The tests found that safeguards varied across models and could be bypassed by switching tools or changing versions.

• Stanford AI narrows 1.7 million polymers to 10 antibiotic leads

  Stanford researchers trained an AI model to search 1.7 million polymer candidates for molecules that mimic bacteria-killing peptides. Ten candidates outperformed expectations against E. coli, with one showing particular promise against biofilms.