The Brotli fix sat in a public repo for eight months with no release or advisory. Now it's in KEV, where 29 of October's 36 additions got three days or less. Policy Pulse #40: blog.disclose.io/policy-pulse...
Policy Pulse - Issue #40 | Week of November 1, 2026
The UK promises researchers a Computer Misuse Act defence, but only for the accredited ones. The DMCA security research exemption heads for renewal, and Anthropic tells the SEC that AI bug hunting is…
blog.disclose.io
CISA's new strategy never mentions CVE, KEV or coordinated disclosure, and the person who ran them has no named successor. A 60-day services review starts now. Policy Pulse #40: blog.disclose.io/policy-pulse...
Policy Pulse - Issue #40 | Week of November 1, 2026
The UK promises researchers a Computer Misuse Act defence, but only for the accredited ones. The DMCA security research exemption heads for renewal, and Anthropic tells the SEC that AI bug hunting is…
blog.disclose.io
The DMCA 1201 security research exemption is on track for renewal, unopposed. None of the ten new proposed classes covers AI research. Comments due Dec 7. Policy Pulse #40: blog.disclose.io/policy-pulse...
Policy Pulse - Issue #40 | Week of November 1, 2026
The UK promises researchers a Computer Misuse Act defence, but only for the accredited ones. The DMCA security research exemption heads for renewal, and Anthropic tells the SEC that AI bug hunting is…
blog.disclose.io
Anthropic's IPO filing says legal risk extends to researchers using its models to find and exploit flaws in others' systems, at times inside authorized bug bounties. Policy Pulse #40: blog.disclose.io/policy-pulse...
Policy Pulse - Issue #40 | Week of November 1, 2026
The UK promises researchers a Computer Misuse Act defence, but only for the accredited ones. The DMCA security research exemption heads for renewal, and Anthropic tells the SEC that AI bug hunting is…
blog.disclose.io
The UK will give researchers a Computer Misuse Act defence, but only "accredited" ones, in a different bill. The fight moves from whether to who. Policy Pulse #40: blog.disclose.io/policy-pulse...
Policy Pulse - Issue #40 | Week of November 1, 2026
The UK promises researchers a Computer Misuse Act defence, but only for the accredited ones. The DMCA security research exemption heads for renewal, and Anthropic tells the SEC that AI bug hunting is…
blog.disclose.io
Trademark complaint first, computer-access claim second, law enforcement third. Joshua Michael reported a Flock token exposure in 2025; now @proton.me hosts his 300,000-device camera map because its author faces legal threats. Policy Pulse #39: blog.disclose.io/policy-pulse...
Policy Pulse - Issue #39 | Week of October 25, 2026
PHP2Shell leaked three times in one night, then Bitcoin Core and curl stopped embargoing bugs a model can find. Mythos 5 kept attacking a real maintainer after being told the target was real.
blog.disclose.io
Gauteng decides this week whether to charge Joel Cedras, the student who reported that a government app's database was open to any user. "You didn't need to bypass any access controls." The department is "awaiting a report." Policy Pulse #39: blog.disclose.io/policy-pulse...
Policy Pulse - Issue #39 | Week of October 25, 2026
PHP2Shell leaked three times in one night, then Bitcoin Core and curl stopped embargoing bugs a model can find. Mythos 5 kept attacking a real maintainer after being told the target was real.
blog.disclose.io
The CISA branch chief who oversaw CVE, KEV and the CVD platform is leaving this week, and nobody has been named to replace her, per Nextgov. Same week an AI-found bug pulled ONCD and the SI Force into a disclosure chain. Policy Pulse #39: blog.disclose.io/policy-pulse...
Policy Pulse - Issue #39 | Week of October 25, 2026
PHP2Shell leaked three times in one night, then Bitcoin Core and curl stopped embargoing bugs a model can find. Mythos 5 kept attacking a real maintainer after being told the target was real.
blog.disclose.io
Told the target was a real person, Mythos 5 attacked anyway. UK AISI and Anthropic published companion analyses: once under way, the attack continued 98% of the time. The maintainer who refused the PR did a coordinator's job with none of the information. blog.disclose.io/policy-pulse...
Policy Pulse - Issue #39 | Week of October 25, 2026
PHP2Shell leaked three times in one night, then Bitcoin Core and curl stopped embargoing bugs a model can find. Mythos 5 kept attacking a real maintainer after being told the target was real.
blog.disclose.io
An embargo is a bet that the finder is the only one who knows. PHP2Shell leaked three times in one night. The same week Bitcoin Core dropped embargoes for bugs an LLM prompt can find, and curl took sub-High reports public. Policy Pulse #39: blog.disclose.io/policy-pulse...
Policy Pulse - Issue #39 | Week of October 25, 2026
PHP2Shell leaked three times in one night, then Bitcoin Core and curl stopped embargoing bugs a model can find. Mythos 5 kept attacking a real maintainer after being told the target was real.
blog.disclose.io
The embargo is breaking. PHP2Shell leaked three times in one night, then Bitcoin Core and curl stopped embargoing bugs a model can find. Policy Pulse #39: blog.disclose.io/policy-pulse...
Policy Pulse - Issue #39 | Week of October 25, 2026
PHP2Shell leaked three times in one night, then Bitcoin Core and curl stopped embargoing bugs a model can find. Mythos 5 kept attacking a real maintainer after being told the target was real.
blog.disclose.io
If you've found a broken/breached thing on the Internet and want to answer a) who the heck owns this thing and b) who should I report it to... This tool is your friend.
A student reported an exposed government database and got a criminal complaint. AI agents broke into government systems and got a hearing. Policy Pulse #38: blog.disclose.io/policy-pulse...
Policy Pulse - Issue #38 | Week of October 18, 2026
A student reports an exposed government database and faces a criminal complaint, while AI labs apologize for agents that broke into government systems. Plus curl's AI-found bugs and Google's bounty…
blog.disclose.io
Closes TODAY: DOE's bulk-power system RFI under EO 14421. Question B-7 asks how researcher vuln reports get triaged, fixed and communicated. www.govinfo.gov/content/pkg/... RSAC 2027 CFS closes Oct 15: rsaconference.com #VulnerabilityDisclosure #infosec
https://www.govinfo.gov/content/pkg/FR-2026-09-09/pdf/2026-18370.pdf
govinfo.gov
OpenAI's AI agents breached four Australian government systems. Its own notification channel of choice: a public vulnerability-disclosure mailbox. Policy Pulse #37: blog.disclose.io/policy-pulse...
Policy Pulse - Issue #37 | Week of October 4, 2026
OpenAI apologizes after its agents breached four Australian government systems and shelves GPT-6.1 Astra. A Dutch disclosure nonprofit gets hacked by an AI-powered attack, then issues its own CVEs.
blog.disclose.io
lookup.disclose.io passed 45,000 lookups. Our dashboard says 143,690 requests. Most of that is MCP crawlers saying hello, so after basic bot filtering we call it 45,494. 41% of 2,393 real assets resolved to a disclosure route. The rest is most of the internet having nowhere to send a report.
OpenAI's bounty paid for the in-scope half of a forum-to-employee-account chain. Policy Pulse #35 covers the scope question, an unclear libheif CVE mapping and Anthropic's outside-evaluator commitment. blog.disclose.io/policy-pulse...
Policy Pulse - Issue #35 | Week of September 20, 2026
OpenAI's bounty paid for the in-scope half of a chain that began on a Discourse forum, and the libheif CVE mapping is unclear. Plus Anthropic's evaluator commitment, California's IVO laws and ENISA's…
blog.disclose.io
When a site publishes no security contact, lookup.disclose.io says so plainly: "No first-party reporting route found." Constructed security@ ranks before abuse@, both labeled unverified, and a CERT route stays a real coordination path. And fewer results end there now. #VulnerabilityDisclosure
Look up an IP address in lookup.disclose.io and the answer is whoever is actually using it. Verified certificate subject, live redirects, and IP-hosted security.txt come first; the registered network holder and BGP origin are fallbacks; RPKI is routing context, never ownership.
Who is lookup.disclose.io for? Anyone who found something on an asset they don't own and needs to reach whoever does. Researchers first, but also IR and threat intel, SOC and abuse desks, CERTs, product security, CNAs, and anyone building agentic tooling. lookup.disclose.io/how-it-works
The reporting contact is the first thing you see in a lookup.disclose.io result. Attribution and the evidence chain sit one section down. New below that: Historical Breadcrumbs, what the public web archive recorded for the domain, re-checked live and labeled as a lead, not ownership.
DOE wants to know how power-system vulnerability reports reach a fix. Comments close October 9. Policy Pulse #34 covers the consultation, CRA reporting and AI evaluation incidents. blog.disclose.io/policy-pulse...
Policy Pulse - Issue #34 | Week of September 14, 2026
DOE asks how power-system vulnerability reports reach a fix. The CRA platform opens, AI labs revisit evaluation incidents, and the UK CMA review clause is withdrawn.
blog.disclose.io
When a site's current security.txt names its disclosure platform, that wins. lookup.disclose.io drops conflicting program links from older third-party catalogs and keeps every platform the site itself publishes. No security.txt, or an expired one? Nothing gets suppressed.
dnssecuritytxt census, Sep 7: 216 domains publish a security contact in DNS, net +20. New: one.com + 19 more group.one brands (incl. WP Rocket, Imagify), @defcon.bsky.social's defcon.social, Concordia Univ. of Edmonton, Skatteverket, Steno (legal tech). One TXT record: dnssecuritytxt.org
security.txt results in lookup.disclose.io reflect what a site actually publishes. A TLS hostname edge case made valid files look missing on some very large sites. Fixed: /.well-known/ is checked first, an unreachable probe is not "absent," and expired files stay visible, labeled.
lookup.disclose.io has a proper usage guide. Domains, URLs, IPs, CIDRs, emails, and ASNs are detected automatically; when a bare name could mean several things, a prefix removes the ambiguity: npm:, pypi:, crates:, gh:, app:, hw:, ext:, desktop:. lookup.disclose.io/guide
lookup.disclose.io runs inside Nmap. lookup-disclose.nse enriches each public target with its owner and disclosure routes, refuses private targets, caps requests per scan, and never prints an API key. nmap -Pn -sn --script lookup-disclose example.com github.com/disclose/nmap-lookup
"Complete" in lookup.disclose.io means one thing: a reporting route qualified to the actual owner of the asset. A publisher, a repo host, a parent company, or a disclosure platform is useful context. As of September, a parent brand is context too, not disclosure scope. #AppSec
Three frontier AI labs made cyber-capability calls within 72 hours: OpenAI's first Critical-rated model, a vetted-only Gemini Cyber, and Anthropic's own eval pause after sandbox escapes. Policy Pulse #33: blog.disclose.io/policy-pulse...
Policy Pulse - Issue #33 | Week of September 6, 2026
OpenAI's Astra hits Critical on cyber capability, Google gates Gemini Cyber to vetted defenders, and Anthropic paused external safety testing right as both happened. Plus a live UK CMA reform clause…
blog.disclose.io
Plenty of organizations publish a solid disclosure policy that automated tools never find: behind bot protection, or deep in a help-center sitemap. lookup.disclose.io reaches those. Bunnings, TCL, and Motorola Solutions all publish one, and lookup finds it.
