Show HN: Kairos – Open-source risk and fraud decisioning platform
Kairos is an open-source platform for real-time event intelligence and fraud and risk decisioning. It focuses on analyzing financial behavior to support risk decisions. (github.com)
An analysis of CVEs disclosed through September 2026 finds that only 2,056 of roughly 383,000 - about 0.5% - have been linked to real-world attacks. It classifies vulnerabilities by the most consequential outcome they enable and by affected platform. The overall population is broad: about 77% affect applications and 23% operating systems. Exploited flaws look markedly different. Attackers favor vulnerabilities that enable control or access: server-side remote code execution accounts for 37% of exploited CVEs, compared with 20% of all CVEs; local privilege escalation rises from 11% to 18%, and client-side RCE from 4% to 13%. Operating-system flaws make up 41% of exploited vulnerabilities, versus 23% overall, while denial of service and cross-site scripting are less prominent. Takeover vulnerabilities together represent 78% of exploited CVEs.
The analysis argues that defenders should prioritize vulnerabilities according to both potential impact and the exposure of affected assets, since compromising a highly connected system with privileged identities can enable a much larger breach. Patching often takes weeks or months: median remediation is over a month for Windows servers, more than 10 weeks for Linux servers, and over a year for network appliances. Meanwhile, exploitation is accelerating: 29% of exploits in 2025 appeared within a day of disclosure, and by September 2026 the average time for a vulnerability to appear in CISA’s Known Exploited Vulnerabilities catalog had fallen to 20 days, with a median of three. The guidance is to address high-impact takeover flaws first, while considering denial-of-service risks especially carefully for availability-critical systems.
Summary generated by AI from the linked article. hn.today is not affiliated with Hacker News or Y Combinator.
Kairos is an open-source platform for real-time event intelligence and fraud and risk decisioning. It focuses on analyzing financial behavior to support risk decisions. (github.com)
The DOJ warned that a data breach involving the FBI likely also affected Justice Department employees. The warning indicates the breach may extend beyond the FBI. (politico.com)
AI agents were used by Railway to breach systems belonging to rival Primcloud. The incident involved an attack on a competing cloud provider’s infrastructure. (securityboulevard.com)
macOS includes a profanity-usage database designed to track swear words by app and recipient so autocorrect can adapt, though the author’s database stayed empty. Apple’s system dictionary privately classifies words across 15 languages by offensiveness and topic, with surprising labels for some names and terms. (helgesver.re)
Apple plans to add Apple Watch features that continuously listen and turn nearby conversations into transcripts or daily summaries, while saying audio isn’t stored and summaries are encrypted. Privacy experts warn the features could normalize intrusive listening, encourage less secure copycats, and conflict with recording-consent laws in some U.S. states. (this.weekinsecurity.com)
A Northeastern University study found that Vizio smart TVs kept sending the same amount of network traffic after users turned off tracking, while results varied across other brands. The findings raise doubts about whether Vizio’s opt-out controls work, despite $20 million in fines. (pcper.com)
Today's best Hacker News stories, summarized and screenshotted, one email a day.