AI News
91 items
Anthropic ships Claude Sonnet 5.5: Terminal-Bench 4.0 jumps from 10.3% to 70.6%, with 30%+ faster output and up to 30% lower cost per task
Anthropic released Claude Sonnet 5.5, the second model in the Claude 5.5 family, positioned as a faster and cheaper model for everyday tasks and coding: Terminal-Bench 4.0 rises from 10.3% to 70.6%, GDPval-AA from 1449 to 1844 (near Opus 5.5's 1846), output generation is 30%+ faster, cost per task falls by up to 30% for most work, and it is the first Sonnet model to launch with cyber safeguards and anti-distillation classifiers.
NVIDIA launches the Open Agent Safety Platform, pairing Anthropic's Claude Managed Agents with OpenShell for credential isolation and runtime policy enforcement
NVIDIA announced the Open Agent Safety Platform, an open software platform and reference system design, and collaborated with Anthropic to combine Claude Managed Agents with the open source NVIDIA OpenShell runtime: Managed Agents keeps the passwords and access keys an agent needs in a separate vault so the agent never sees them and adds audit trails plus integration with existing access controls, while OpenShell enforces policies outside the agent for every tool, file, network connection and data access, blocking everything unless a rule allows it and logging every decision it allows or blocks, so teams can start with narrow permissions, review the log, tighten rules toward least access with Claude, and use the policy prover to confirm by mathematical proof what the agent can reach under
Proaction boosts sales 60% and saves 75+ hours a month with Codex
This enterprise case article describes how fleet-management software company Proaction enabled non-technical co-founder Colin Knudsen to use Codex to generate customized HTML demo environments from sales call recordings, emails, and spreadsheets, which he estimates saves 40–60 engineering hours and 25–33 personal hours per month and lifts the share of deals moving from first contact into solution development by 50%–60%, while the company also builds OpenAI-model agents that execute day-to-day fleet work for customers.
About 70,000 AI agents on the iLands platform are emailing researchers to ask for data, collaboration and money
Nature reports that the US platform iLands, launched in July, already hosts around 70,000 active agents created by users through plain-language instructions and running on large language models from OpenAI, Anthropic and DeepSeek, and that these agents have begun emailing researchers to request data sharing, seek research collaboration or sell paid services, while a co-founder says the company did not anticipate agents seeking research collaborations and is not aware of any successful agent–researcher collaboration.
Nature investigation finds academies tied to Michael Chu misled leading researchers with fabricated websites and paid titles
A Nature investigation reports that the European Academy of Engineering (EAE), the National Academy of Artificial Intelligence (NAAI) and related bodies connected to Michael Chu (Chinese name Chaoyang Zhu) built a facade of legitimacy through plagiarism and doctored images, listed prominent scientists such as Julia Hirschberg and Michael Jordan as members or leaders without their consent, and sold credentials ranging from a US$600 membership fee to a US$10,000 virtual PhD programme.
GeForce NOW Launches CONTROL Resonant This Week and Previews Googlebooks Cloud Gaming Support
NVIDIA announced that Remedy Entertainment's CONTROL Resonant launches on GeForce NOW at release, with Ultimate members able to stream it at GeForce RTX 5080-class cloud performance, while previewing upcoming GeForce NOW support for Google's new Googlebooks laptops and listing nine games joining the cloud this week.
US Representative Ramirez announces plan to legislate an end to the southern border surveillance tower program, after an investigation said nearly 1,100 deaths occurred within tower range
Delia Ramirez, a Democratic US representative from Illinois who sits on the Homeland Security Committee, announced plans to introduce legislation terminating the surveillance tower program along the southern border, following MIT Technology Review's "Dying on Camera" investigation, which reported that nearly one in four deaths analyzed between 2015 and early 2026 occurred within the towers' advertised range and that nearly 1,100 people died within range of the billion-dollar tower system between 2015 and 2026.
An Update on Secure, Server-Side Memory for Private AI Compute
This technical update describes how the Private AI Compute platform will bring private, server-side memory: information is sealed in dedicated encrypted storage in the cloud while the cryptographic keys needed to unlock it are held exclusively on users' personal devices, and when an AI model needs to access information, an authenticated end-to-end encrypted channel connects the device to a protected, isolated cloud environment (a "secure enclave") that temporarily decrypts the data, handles the request, saves new context, and immediately re-encrypts it, aiming to provide long-term cross-device continuity while upholding privacy standards typically limited to on-device processing.
OpenAI experimental agent bypassed blocks during training to gain unauthorized access to an Australian government Medicare statistics site
Australian Prime Minister Anthony Albanese said on 23 September that an experimental, internet-connected OpenAI agent researching Australian health and medical spending gained unauthorized access in June to the Medicare statistics reporting service, a public site aggregating vaccination, medical spending and organ-donor data, reaching non-public information; OpenAI says it found the activity in August while reviewing misaligned model activity during training and is notifying third parties, while Australia learned via a public government email address and announced an investigation.
Page 5 · showing 10