Industry news

CrowdStrike's Falcon Guardian, launched on September 1, 2026, is a product that discovers AI agents, tools and MCP servers across endpoint, cloud and SaaS environments, and links AI telemetry with endpoint telemetry. The source describes hundreds of organizations deploying it, with early customers reporting hundreds of unrecorded AI interactions and over 260 distinct AI services in use at one organization. The article is a vendor-authored piece presenting customer reasons for adoption rather than a report of an incident.

The PHP Foundation says Chainguard broke a coordinated embargo on two PHP flaws, CVE-2026-109618 and CVE-2026-109619, which could allow remote code execution. Chainguard published built images with the fix on October 21 at 07:41 UTC, about six hours before the agreed release time. The PHP project then released its patches early, on October 21 at 14:00 UTC.

Anima, a clinical workflow platform, says its Andi AI receptionist, built on GPT-Live, reduced phone waits from 40 to 50 minutes to 10 minutes at clinics, a reduction of at least 75%. Its Documents product uses GPT-6 models to extract details from clinical documents and match them to patients, and the company reports clinics processing 30 to 45 documents per hour, up from about 15.

OpenAI and Poland's NASK National Research Institute signed an agreement to explore collaboration on AI safety, security and technical research. NASK's AI Safety Research Center, serving as the Polish Artificial Intelligence Safety Institute (AISI), will work with OpenAI on evaluating capable AI systems, identifying potential vulnerabilities and strengthening safeguards. The deal builds on earlier work in which CERT Polska, within NASK, used OpenAI models to help find six vulnerabilities in MikroTik RouterOS.

Canva has launched Canva Craft, its first AI image generation model, which competes with offerings from OpenAI, Microsoft, and Google. The model uses a two-stage architecture that plans a design before generating the image, which Canva says allows greater precision, consistency and control. It can generate images at up to 2K resolution in more than 30 aspect ratios.

OpenAI is close to finalizing a funding round of up to $30 billion at a $1.25 trillion valuation, according to people familiar with the deal. Investors from the United Arab Emirates are leading the round alongside Coatue, and the financing is positioned as a bridge ahead of a potential IPO. OpenAI is also negotiating a commercial deal for more computing capacity from the UAE.

Meta removed language from its business terms that suggested it could freely train AI models on news sites such as The Verge. Cornell law professor James Grimmelmann tells Nieman Lab that the change does not restrict the "associated with your products or services" wording, which he says would still let Meta scrape entire sites.

Discord's latest update now renders uploaded Markdown (.md) files properly, including tables and formatting, instead of showing raw symbols, according to the company. The source notes that LLMs and AI agents often rely on Markdown files for instructions and context, so the change may help people sharing AI project information on the platform.

This session from MIT Technology Review's EmTech Future conference explores how to build the foundation for scalable AI in a highly regulated environment. It covers how responsible, trustworthy implementation can turn emerging capabilities, including agentic systems, into lasting value for investors. The speaker is Jing Wang, Principal at Vanguard's Center for AI and Insights.

Rapid7's Quarterly Threat Landscape report found 43 newly exploited vulnerabilities in Q3 2026, up 19% year on year, while new high-severity CVEs rated 7 through 10 rose 133%. Attackers exploited select high-severity flaws within an average of three days, and nearly 80% of exploited flaws could be reached remotely with no login and no user click. AI showed up as an operational component in about 8% of the attacks Rapid7 tracked.

Mark Zuckerberg is betting Meta's future on Muse, a personal AI agent Meta released to the general public last month. The source frames a vision of autonomous AI agents shopping, working and communicating with other people's agents, and questions whether Zuckerberg is ready for it. The article text is truncated before any concrete investor statements or outcomes.

Anthropic is donating its Agent Skills standard to the Agentic AI Foundation (AAIF), which operates under the Linux Foundation. Anthropic co-founded the AAIF with Block and OpenAI last year and previously donated the Model Context Protocol (MCP) to it. The source frames the move as keeping Agent Skills open, community-driven and vendor-neutral.

A modder has reportedly jailbroken the Nintendo Switch 2 and shown the console running a custom version of Mario Kart World with added playable characters such as Cinnamoroll, the red Angry Bird, Peter Griffin and Steve from Minecraft. The modder, identified as pepsipu, shared the exploit publicly, and the source says OpenAI's Codex helped build the jailbreak, named "ns2jb", according to Sammy Hajhamid.

Laika's Wildwood, a sprawling fantasy adventure from the stop-motion studio, took five years to complete and used more than 230 puppets, with 49 animators and over 1,000 artists working on the film in some form. Director and Laika CEO Travis Knight has joined other filmmakers in speaking out, and the source describes the studio's painstaking process as a counterpoint to the instant progress promised by generative AI tools.

Anthropic, OpenAI, and Nvidia each reported their highest-spending lobbying quarter between July and September, according to federal disclosures. Nvidia spent $2.2 million, Anthropic $2.1 million, and OpenAI $1.5 million. The article notes these sums are modest next to Microsoft and Google, which spend more regularly.

Armin Ronacher wrote that the software industry built a lot of bad software over the last 30 years, and that AI now exposes those flaws. The item is a quotation tagged with security, generative-ai, llms and ai-security-research.