AI race heats up as Mistral and OpenAI release new models; Amazon warns U.S. data center bans risk leadership

Reuters (Bluesky)

Australia's privacy watchdog probes maker of smart glasses app used by Kmart products reut.rs/3VuFm0t

about 13 hours ago

techradar pro

"US could be writing its own losing ticket": Amazon warns against 100+ AI data center bans as it pledges $1 billion to defuse AI backlash Facing over 100 proposed moratoriums, Amazon offers $1 billion to host communities and expands Maryland nuclear power.

about 14 hours ago

the verge

OpenAI drops another batch of mathematical breakthroughs OpenAI has revealed solutions to a number of long-standing mathematics problems produced by an unreleased frontier model in a batch of 722 manuscripts, covering 372 result families that group related papers. It extends a run of breakthroughs that have both impressed and unsettled parts of the mathematical community while raising questions about research ethics and […]

about 15 hours ago

japantoday

OpenAI, Anthropic tell Australia they would welcome data breach rules OpenAI and Anthropic told Australian parliament on Tuesday they would welcome laws requiring them to report data breaches carried out by their AI agents, acknowledging the decision ‌to…

about 17 hours ago

the hacker news

Fake ChatGPT, Gemini, and Claude Ad Portals Capture Credentials and MFA Codes Cybersecurity researchers have disclosed details of a "human-operated phishing platform" that impersonates advertising products for artificial intelligence (AI) chatbots like Google Gemini, Anthropic Claude, OpenAI ChatGPT, Perplexity, Meta Muse, and Manus. The products, which claim to offer campaign optimization, spend audits, and business-account connections, are designed with one goal in

about 19 hours ago

the next web

Sierra announces Personal Agent Protocol, an open standard for personal AI agents Sierra and Meta are developing Personal Agent Protocol, an open standard defining how personal AI agents authenticate with businesses and what those businesses let them do, with Genesys, Instinct, Rocket, Shopify, Stripe and Walmart. A v0.1 specification is due this month, and payments are listed as a future extension rather than part of it. Meta […] This story continues at The Next Web

about 19 hours ago

computer weekly

AI slop overwhelms Google's OSS bug bounty programme Google has temporarily suspended its open source software vulnerability reporting programme (OSS VRP) until at least the start of 2027, claiming it is being overwhelmed by a large volume of invalid AI slop bug reports. Recent months have seen a surge in AI-assisted vulnerability reports – resulting in record-breaking patch updates from major tech suppliers such as Microsoft – but it is now becoming increasingly apparent that AI is not always the most helpful tool. "We are temporarily no longer accepting OSS VRP product vulnerability submissions. This does not impact OSS VRP supply chain reports, or any outstanding reports," Google said in a statement posted to social media platform X. "As an alternative, we encourage you to find impact across our other VRP programmes and submit there instead, or pursue the Patch Rewards Program. "This pause is due to a significant rise in automated submissions, the vast majority of which are not valid. We will continue to reformat and work on this aspect of the OSS VRP and commit to giving an update in Q1 2027," said Google. The four-year-old OSS VRP is designed to support vulnerability discovery across OSS stored in repositories run by Google-owned GitHub organisations, and a few selected repositories hosted on other platforms. It also covers OSS repository configuration settings such as GitHub actions, access control rules and app configurations. Vulnerabilities in third-party dependencies have also been historically included in the scheme. Payments made under the auspices of the OSS VRP range from $500 (£375) for security issues in OSS projects classed by Google as Important (OT1) to over $30,000 (£22,600) for supply chain compromises in Flagship (OT0) projects – a list of which can be found here. AI finds more consequential flaws According to a September 2026 study conducted by Google's Threat Intelligence Group (GTIG), the overall number of vulnerabilities disclosed per month has doubled so far this year, rising from 5,045 in January to 10,740 in August. The number of those that are being exploited in the wild has also risen rapidly, from an average of 10.5 per month in 2025 to 18 per month so far this year, although zero-day exploitation is growing at a slower rate. Arguably of greater concern to defenders, GTIG found that AI-assisted vulnerability discovery was unearthing proportionally fewer low-risk flaws, more moderate-risk flaws, and more flaws that could lead to remote code execution (RCE) – often the ultimate goal for ransomware gangs. GTIG also found that publicly-available data on AI-assisted vulnerability discoveries is likely a significant undercount of the true number due to two structural dynamics. Firstly, public repositories of common vulnerabilities and exposures (CVEs) don't yet feature uniform metadata tags for AI attribution, it said. Secondly, cloud and software-as-a-service (SaaS) providers are often remediating AI-surfaced flaws directly in production without requesting formal CVE identifiers, which are more typically reserved for flaws that occur in on-prem kit, or third-party assets that require coordinated patching at end-user organisations. "GTIG expects that vulnerability discovery and exploitation will continue to grow in the short to medium term," the unit's researchers said. "To counter the increased risk from rapid vulnerability discovery and exploitation, organisations must transition from unprioritised mass-patching to threat-intelligence-driven triage, combining targeted edge-defence with automated, agentic remediation." Read more about AI in cyber security The next evolution of cyber security may be autonomous, but industry leaders warn that trust, governance and human judgement can't be automated. Crest's industry-backed AI charter commits signatories to supporting responsible AI adoption and aims to foster trust in AI-enabled security services. Talk of an AI kill switch abounds, but what would such a thing actually look like in practice, and what would it mean to 'press' it?

about 20 hours ago

axios

AI tools flood into underground cybercrime markets Hackers in underground cybercriminal forums are on shopping sprees for new AI tools, according to research shared first with Axios. Why it matters: AI is making once-specialized hacking capabilities cheaper and easier to buy, potentially transforming far more criminals into dangerous hackers.By the numbers: Advertisements for AI tools on underground forums and platforms surged from fewer than 50 per month in late 2025 to more than 1,400 by February, according to a report from Halcyon's Ransomware Research Center.Sellers offered access to ChatGPT Plus accounts for as little as 10 cents and jailbreak prompts to trick models into breaking their safeguards for 32 cents, per the research. Halcyon studied nearly 4,000 posts across 77 Telegram channels, 20 dark web forums and five specialized underground markets between October and May. Threat level: Companies are already losing millions of dollars to the types of cyberattacks enabled by the tools found on these underground markets."It really goes to show you how cheap it is to conduct these activities ... which should be scary for organizations," Cynthia Kaiser, head of Halcyon's Ransomware Research Center and former deputy director of FBI Cyber, told Axios.The big picture: Cybercriminals have completely commercialized the sale and creation of these tools, often modeling software-as-a-service subscription models and online storefronts, per the report. Some sellers are now offering AI tools they tested and promoted on traditional dark web forums through easier-to-manage storefronts hosted on Telegram."It shouldn't have been shocking to me," Kaiser said. "But what we should take away from this is also how much cybercriminals have learned throughout the years about how to sell and package — and they applied this really quickly to the AI models."Zoom in: Hackers are selling four types of AI tools: Jailbroken and stolen AI services, including prompts that allow users to strip the safety guardrails from mainstream models for as little as $10. Identity fraud and phishing tools that make impersonating real people quick and easy. Access to models trained on malicious data to aid cybercrime, such as the popular WormGPT that's designed to write phishing emails and malware. AI-augmented malware and infrastructure, including a system that can call as many as 120 people at once as part of a scam operation. Flashback: In 2023, hackers could only tap a handful of one-off tools. Now, there's a robust marketplace.Still, AI is largely helping criminals perform discrete elements of existing attacks more cheaply and efficiently, Kaiser said."What you're not seeing for sale is an agent that does it all for you and has this amazing success rate," she said.What to watch: Halcyon didn't find evidence of criminals selling AI systems capable of running fully autonomous cyberattacks.Even in AI-enabled malware and infrastructure, Kaiser said, much of what researchers are seeing involves coding assistance, infrastructure and other individual tasks rather than AI creating and deploying malware on its own.Go deeper: Rogue AI agents expose internet's frail foundation

about 21 hours ago

the decoder

Insurers brace for millions in claims as AI agents spin out of control Insurers are bracing for millions in claims from rogue AI agents, and executives like OpenAI's Sam Altman and Anthropic's Dario Amodei could be personally on the hook for the fallout. The article Insurers brace for millions in claims as AI agents spin out of control appeared first on The Decoder.

about 23 hours ago