Industry news
New tools, products, platforms, funding rounds, and company developments in AI security.
New tools, products, platforms, funding rounds, and company developments in AI security.
6,034 items
The Guardian is asking readers how they have used AI to help with online dating, such as writing chat-up lines or getting social coaching. Respondents are invited to describe what worked and what did not, through an encrypted form that accepts anonymous submissions.
Matthew Green has published a site estimating how much public cryptanalysis each public-key assumption has received, using about sixteen thousand candidate papers. The study covers ECDLP, lattices, and syndrome decoding, converting papers into estimated research-allocation years. The author notes that publication-based estimates may substantially overstate actual time spent, and that unpublished work, including secret intelligence-agency cryptanalysis, is not counted.
A growing set of consumer gadgets collects data about people around the user without their consent. Meta's Muse, an AI personal assistant released last month, asks users to share information about friends, family and other contacts. The agent has reportedly been downloaded millions of times and reached No. 1 on Apple's App Store.
Anthropic has launched OSS Scanner, a free, opt-in service that uses its Claude models, including Mythos, to find vulnerabilities in open source projects. Participating maintainers receive reproduction details and proposed fixes when available, without waiting for Anthropic's researchers to validate the findings. Over six months, Anthropic reports 29,000 candidate vulnerabilities, of which about 6,000 were manually reviewed and triaged.
AWS, Anthropic, OpenAI and Google are racing to build forward deployed engineering teams, as companies seek returns on AI investments. Job postings for these engineers rose from fewer than 200 in 2023 to more than 7,300 in the first nine months of 2026, with median advertised salaries reaching $195,456 according to Lightcast.
A viral Instagram post attributed to actress Mia Goth, showing her in a T-shirt reading "MY EX IS MY BIGGEST FAN" with a Beyoncé song, circulated after she reportedly filed for divorce from Shia LaBeouf. The post came from the account @miagoth_, which had over a million followers and a blue verified checkmark, yet the image was fake, and it spread within days.
Intruder's AI pentester is compared with human pentesters in an article drawn from the author's twelve years of pentesting experience. The author says the AI agents are given broadly defined roles and freedom, with stricter guardrails to prevent damage, rather than a strict methodology, and this produced an authentication bypass that had gone unnoticed through five years of human pentests.
Musician, YouTuber and investigative journalist Benn Jordan has become a prominent critic of Flock Safety, maker of automatic license plate readers (ALPRs). He testified at a Senate Judiciary Committee subcommittee hearing on Flock and other ALPR systems. The source says hundreds of local governments have canceled Flock contracts, Sen. Josh Hawley is investigating the company, and Reuters reported Flock plans to cut about 18 percent of its workforce.
ARMORY CASTLE, a suspected PRC-nexus espionage actor also tracked as UNC5221, exploited CVE-2026-102764 as a zero-day in late August 2026 against BeyondTrust Remote Support appliances, according to Mandiant. After exploitation, the actor deployed the SAWFLY PHP webshell and the BARECLAW tunneler, escalated privileges, and deleted logs to maintain access and evade detection.
Fix: BeyondTrust provides security patches to remediate the vulnerability, and Mandiant strongly recommends that affected organizations apply them immediately.
Google Threat IntelligenceAnthropic told investors that quarterly revenue rose nearly 50% to $16.8bn in the third quarter, according to people with knowledge of the matter reported by the FT. The company said distillation, where rival labs train models on Anthropic's outputs, accounted for 5 to 7 per cent of revenue and was cut back materially in Q3.
Google has begun rolling out Universal Cart in the US across the Gemini app and Google Search. The tool, previewed at Google I/O in May, is a single digital cart that saves products from multiple retailers and works across Google platforms, with the aim of helping users stay organized and save money ahead of Black Friday and the holidays.
Attackers impersonate AI products such as Claude, ChatGPT, GitHub Copilot and Gemini to distribute malware, steal browser sessions and compromise business accounts. ANY.RUN linked 1,258 public sandbox analyses to fake AI campaigns between April and October of 2026, and US businesses accounted for 16% of country-attributed targets. The campaigns add operational strain on MSSPs, since similar attacks across customers drive repetitive investigations and longer response times.
A Semperis 2026 Ransomware Risk Report surveyed 1,200 IT and security professionals across nine countries and eight sectors. It found that 84% of ransomware victims faced attacks involving AI, 87% reported attacks on identity infrastructure, and 37% said employees' friends or family had been threatened. Attackers are extending extortion beyond corporate networks to employees' personal lives.
Google Threat Intelligence Group and Mandiant investigated a supply chain compromise targeting AOMEI Technology's update infrastructure between April and July 2026. The threat actor, tracked as UNC7017, repeatedly tampered with AOMEI origin server configurations for five to ten minutes at a time, delivering a multi-stage payload chain involving the CROSSROAD downloader and SHINYROAD backdoor.
Fix: AOMEI reported that they took corrective action to secure their distribution servers.
Google Threat IntelligenceThreat actors are exploiting CVE-2026-105740 (CVSS 9.9) in Langflow, an open-source platform for building AI applications and agentic workflows. The flaw is an authenticated remote code execution issue affecting versions prior to 1.9.0: any authenticated user can add an MCP server with the 'stdio' transport, and its user-supplied command runs via bash -c with no validation. Previdian's honeypots recorded 1,423 exploitation attempts from 27 IP addresses since September 19, 2026, with activity rising from October 11, 2026.
Anthropic's revenue rose 45% to $16.8bn (£12.7bn) in the three months to the end of September, according to figures the BBC understands the company revealed in an investor briefing. The briefing precedes a stock market listing expected before the end of the year, and reports say the company made a profit for the second quarter in a row. Tech and chip stocks have fallen amid investor concern over spending on AI infrastructure.
Snyk introduces Evo Agentic AppSec, a team of security agents that handles application security tasks such as threat assessment, vulnerability detection, prioritization, remediation and verification, with human teams setting policy. The article argues that AI-accelerated development has left security workflows behind, citing roughly eight new vulnerabilities introduced for every one remediated as of October 2026. In Snyk's VulnBump benchmark, the Evo Remediation Agent produced mergeable fixes 90% of the time, compared with 59.6% for the same model without Snyk context.
Snyk published VulnBump, a benchmark of 307 real upgrades of vulnerable dependencies across six package managers (Maven, Gradle, npm, Yarn, pip, Poetry). A task passes only if the vulnerabilities are gone, the project installs, builds, and passes its tests. Snyk's Evo Remediation Agent passed 90.2% of upgrades on Claude Sonnet 5.5, versus 77.5% for Claude Code with Snyk's recommended upgrade and scan and 59.6% for Claude Code unaided.
Anthropic reportedly told investors its third-quarter revenue reached $16.8 billion, up from $11.6 billion in the second quarter and $4.7 billion in the first. The company also said it was adjusted-profitable and cash-flow positive in the third quarter, ahead of an IPO reportedly slated for November.
Simon Willison built a LaTeX renderer inspired by Tiny Little Office, a web app suite that stores document data in the URL #fragment, making each document a shareable link. The tool was built with Claude Code and uses an iframe sandbox, modeled on his SVG-in-Markdown tool, as protection against XSS risks. It loads LaTeX from a URL or gist and uses Pako 2.1.0, MathJax 4.1.3, faster-latex 0.3.1 and DOMPurify 3.4.16.
Fix: Fixed in 1.9.0 (versions prior to 1.9.0 are affected).
The Hacker NewsFix: The source recommends running application security in three places: checks inside the coding workflow, fast deterministic scans on every change in CI, and deeper analysis of the full application at regular intervals. It also recommends clear policies on what gets blocked, fixed or reviewed by a human, and independent validation of proposed fixes before acceptance.
Snyk Blog