AIOpenAI says it is more than quadrupling the midpoint of its medium, high and critical reward ranges and adds quality, loyalty and campaign bonuses.
LanguagesThe 31 October release patches CVE-2026-116238 to CVE-2026-116252, replaces free-text password exec commands with an allowlist and adds an extended-statistics node.
FrameworksThe 1 November release patches path handling in mount(), wildcard gaps in RegExpRouter, a form validator bypass, HTML injection through the css helper and a csrf() gap for hyperlink pings.
ToolingGreg Kroah-Hartman told users to move to 7.3.y, saying keeping 7.2 alive would have meant maintaining eight stable branches at once.
SecurityHacktron's write-up shows how a backtick payload in the account email reached a shell, then superuser, maintainer impersonation, root and a backup server; the FSF fixed 26 reported flaws.
CloudTwo new optional headers, CloudFront-Viewer-ISP and CloudFront-Viewer-TLS-Details, are forwarded to the origin through an origin request policy.
AnalysisNolan Lawson had a coding agent build the same Pokémon app five times per framework; Lighthouse scores barely moved, while build cost and shipped JavaScript differed widely.
News FeedThe company disclosed on 1 November that the exploit is blocked, a subset of organisations had their environment variables exposed and affected owners were emailed.
AIOpenAI says Plus, Pro and Business plans will draw ChatGPT and Codex use from a single allowance measured in tokens; Instant chat will not count against it.
ToolingTeams with a hand-maintained allowlist need to add 9.234.98.160/28, 9.234.98.176/28 and 9.234.106.48/28 or risk rejecting deliveries; allowlists fed from the API need nothing.