Security — Page 6

Axios identified Mistral Large 4 as a 1-trillion-parameter multimodal model with 49 billion active parameters. Mistral says it will release the weights by the end of October.

Lightwell Clearinghouse is now generally available, letting eligible enterprises submit dependencies for priority review and remediation.

At Abu Dhabi's Ai Everything conference, Mistral CEO Arthur Mensch said the model would be unveiled on October 6th; he gave no benchmark, score or rival model names.

JetBrains, Emphere, A-Team Systems and DACHS IT joined the Linux Foundation initiative on October 6th, after manufacturer reporting under the EU Cyber Resilience Act began on September 11th.

The archive operator says it kept a local copy; GitHub's policy allows dual-use research but permits restrictions in cases of widespread abuse.

Logan Graham described the decision at a New York City Council hearing; Anthropic instead routed the model to vetted defenders through Project Glasswing.

JayJay P. says two separate company accounts showed abnormal activity, including one that typically spent $100-$200 a month.

The former CyberArk CEO joins as Keyfactor closes Summit Partners' $1B-plus investment and prepares for its next phase of growth.

Hari Mulackal says Apex can make security-agent work cheaper; Cantina's own held-out test put it just behind Claude Opus 5 High at a fraction of the estimated cost.

Prosecutors allege Earthmade routed servers through Malaysia and Singapore; Lui faces statutory maximums totaling 50 years if convicted on all counts.

Reuters reported that OpenAI was searching roughly 50 petabytes of data; OpenAI's notices cover a range of behavior and do not establish that every recipient was breached.

The July benchmark tested GPT-5.6 Sol Ultra against V8 security fixes; Hacktron's September post recirculated the results, which came from a controlled lab setup.

METR's president used the September 30th hearing to argue for public evidence on agent capabilities, safeguards and incidents, drawing on the July attack on Hugging Face.

Sundar Pichai says the model is already in internal use and will reach developers, enterprises and consumers after a controlled security rollout.

The inspected client code adds four categories to Rampart's default keep-set, while the model card says generic government identifiers have no deterministic backstop.

The move would take Cloudflare from relying on outside certificate authorities to seeking a role in the trust infrastructure behind public HTTPS.

Md Ismail Sojal promoted the Gemma 4 fine-tune on September 29th; its model card reports no formal benchmarks and warns users to verify its conclusions.

GitHub Security Lab says its open-source workflows found and reported 24 flaws. The examples include an OsmAnd location-tracking bug and a Wikipedia app deep-link issue; running the mobile audit requires a Copilot license and premium model requests.

Faav says a missing JWT signature check exposed an analytics environment estimated at 17.3 trillion rows; Microsoft shut the API after his September 5th report.

TheIntelFrog says public Facebook groups surfaced USS Theodore Roosevelt family posts, but its thread does not establish that foreign intelligence ran them.

Researcher Joshua Michael says an unauthenticated token exposed Flock's device inventory; his disclosure dates to November 2025, and the map appeared during a Senate hearing on September 23rd.

ころころゴロリ says Zakynthos flaws could expose protected processes and physical memory; a linked GitHub repository contains C proof-of-concept code.

CEO Saryu Nayyar is extending Gurucul's behavioral-security approach to AI systems; its prevention controls remain in preview.

OX Security reports overseas, home-network and abandoned endpoints among published MCP servers; its findings do not establish enterprise use or compromise.