## Summary
When an application using Pydantic AI opts a URL into `force_download='allow-local'` (which disables the default block on private/internal IPs) **and runs on a network that routes the affected IPv6 transition forms (NAT64- or ISATAP-configured networks)**, the cloud-metadata blocklist could be bypassed by encoding the metadata IP in an IPv6 transition form that the previous fix did not decode — IPv4-compatible IPv6 (`::a.b.c.d`), the NAT64 RFC 8215 local-use prefix (`64:ff9b:1::/48`), operator-chosen NAT64 prefixes, or ISATAP. The IPv6 wrapper is then delivered to the underlying IPv4 metadata endpoint, exposing cloud IAM short-term credentials.
**The bypass is exploitable only in environments whose network actually routes these forms** — NAT64-configured networks (IPv6-only or dual-stack-with-NAT64 deployments, including some Kubernetes setups) for the NAT64 variants, or networks with an ISATAP tunnel for ISATAP. A standard dual-stack cloud VM or container does not route them and is not affected in practice. The IPv4-compatible and Teredo variants are deprecated and addressed as defense-in-depth.
This is an incomplete fix of [GHSA-cqp8-fcvh-x7r3](https://github.com/pydantic/pydantic-ai/security/advisories/GHSA-cqp8-fcvh-x7r3) / [CVE-2026-46678](https://nvd.nist.gov/vuln/detail/CVE-2026-46678) (itself a follow-up to [CVE-2026-25580](https://nvd.nist.gov/vuln/detail/CVE-2026-25580)). The prior remediation decoded only IPv4-mapped IPv6, 6to4, and the NAT64 well-known prefix; the metadata guarantee did not hold for the remaining transition forms.
## Severity
**MEDIUM** — `CVSS:3.1/AV:N/AC:H/PR:N/UI:N/S:C/C:H/I:N/A:N` = **6.8**
Same impact metrics and narrow attack surface as the parent advisory (AC:H): exploitation requires the application to have opted into `allow-local` on a URL influenced by untrusted input, and the NAT64/ISATAP variants additionally require the deployment network to route those forms.
**CWE-918**: Server-Side Request Forgery (SSRF)
## Affected Versions
| Package | Vulnerable | Patched |
|---|---|---|
| `pydantic-ai` | `>= 1.56.0, < 1.102.0`; `>= 2.0.0b1, < 2.0.0b3` | `1.102.0`; `2.0.0b3` |
| `pydantic-ai-slim` | `>= 1.56.0, < 1.102.0`; `>= 2.0.0b1, < 2.0.0b3` | `1.102.0`; `2.0.0b3` |
These transition forms have not been decoded since SSRF protection was introduced in `1.56.0`.
## Who Is Affected
Users are affected **only if** their application explicitly opts a `FileUrl` (`ImageUrl`, `AudioUrl`, `VideoUrl`, `DocumentUrl`) into `force_download='allow-local'` on a URL that is, or could be, influenced by untrusted input.
Beyond that precondition, the affected encodings only reach a metadata endpoint in environments whose network actually routes them. The broadly-routable IPv4-mapped form was addressed in `1.99.0` (CVE-2026-46678); the additional forms addressed here require a **NAT64-configured network** (IPv6-only or dual-stack-with-NAT64 deployments, including some Kubernetes setups) for the NAT64 variants, or an **ISATAP tunnel** for the ISATAP variant. The IPv4-compatible and Teredo forms are deprecated and not routed by modern stacks; they are addressed as defense-in-depth. Most deployments on a standard dual-stack cloud VM or container are therefore not exploitable in practice, but the fix restores the "always blocked" guarantee for the environments that are.
Users are **not** affected if they use any of the bundled integrations to ingest user input, because they do not propagate `force_download` from external data:
- `Agent.to_web` / `clai web`
- `VercelAIAdapter`
- `AGUIAdapter` / `Agent.to_ag_ui`
Applications that only download from developer-controlled URLs are not affected.
## Remediation
Upgrade to `1.102.0` or later (or `2.0.0b3` or later on the 2.0 pre-release line). The cloud-metadata and private-IP blocklists now decode the embedded IPv4 of every standardized IPv6 transition form before evaluating it — IPv4-mapped, IPv4-compatible, 6to4, NAT64 across all prefix lengths (including the RFC 8215 local-use prefix and operator-chosen prefixes), ISATAP, and Teredo. The set of always-blocked cloud metadata/credential endpoints has also been expanded across providers.
## Workaround for Unpatched Versions
Avoid passing `force_download='allow-local'` on any URL that could be influenced by untrusted input. If developers must, resolve the hostname themselves and validate the result against their own metadata blocklist — including IPv6 transition forms — before constructing the `FileUrl`.
## Credits
Reported by [@SnailSploit](https://snailsploit.com).
A user reported issues with the Indian cybercrime complaint website (https://www.cybercrime.gov.in/Webform/Crime_AuthoLogin.aspx), describing problems such as delayed OTP delivery and the website becoming unresponsive during complaint filing. The report originates from a Reddit post with minimal discussion and no technical details or confirmed vulnerabilities. The website currently returns a 'Page Not Found' error.
The US government has ordered Anthropic to block all foreign nationals from accessing Fable 5 and Mythos 5, forcing the company to suspend both models worldwide. Anthropic is complying but disputes the basis, calling the cited jailbreak narrow and the capability widely available elsewhere. [...]
Other noteworthy stories that might have slipped under the radar: US gov targets 72-hour patch cycles, malware uses Windows Phone Link to steal OTPs, spy operation targets Eurasian drone industry. The post In Other News: Train Hacker Arrested, PamDOORa Linux Backdoor, New CISA Director Frontrunner appeared first on SecurityWeek .
Researchers found adware capable of killing cybersecurity products and pushing more dangerous payloads to infected systems. The post $10 Domain Could Have Handed Hackers 25k Endpoints, Including in OT and Gov Networks appeared first on SecurityWeek .
CVE-2026-35581 is a command injection vulnerability in the Executrix utility class of the Emissary project. The vulnerability arises because the PLACE_NAME configuration parameter is sanitized only by replacing spaces with underscores, allowing shell metacharacters to pass through into commands executed via /bin/sh -c. This enables an attacker with the ability to control configuration values to execute arbitrary commands on the host. The issue is fixed in Emissary release 8.39.0 by applying an allowlist regex that replaces all non-alphanumeric, non-underscore, and non-hyphen characters with underscores.
Emissary versions prior to 8.39.0 contain a stored cross-site scripting (XSS) vulnerability in the navigation template. The vulnerability arises because navigation link URLs are interpolated into href attributes without validating the URL scheme, allowing an administrator to inject javascript: URIs. Authenticated users who click such malicious links can execute arbitrary scripts in their browser context, potentially leading to session hijacking and actions performed on their behalf. Exploitation requires administrative access to modify navigation configuration and user interaction. The issue is fixed in version 8.39.0 by adding server-side URL scheme validation and template hardening.
An exposed open directory revealed a comprehensive Roundcube exploitation toolkit used by APT28 to target Ukrainian government entities. The toolkit includes XSS payloads, a Flask-based C2 server, CSS injection tools, and a Go-based implant. It enables credential harvesting, persistent mail forwarding, bulk email exfiltration, address book theft, and 2FA secret extraction. The primary target was identified as mail.dmsu.gov.ua, Ukraine's State Migration Service. Technical analysis shows significant overlaps with previously documented APT28 operations, while introducing new capabilities such as CSS-based side-channel attacks and browser credential theft. The toolkit's modular approach and sophisticated evasion techniques demonstrate APT28's evolving tactics in compromising webmail platforms for long-term intelligence gathering.
Sites running NOAA PMEL Live Access Server (LAS) are vulnerable to remote code execution via specially crafted requests that include PyFerret expressions. By leveraging a SPAWN command, a remote, unauthenticated attacker can execute arbitrary OS commands. Fixed in a version of 'gov.noaa.pmel.tmap.las.filter.RequestInputFilter.java' from 2025-09-24.
Other noteworthy stories that might have slipped under the radar: Jaguar Land Rover sales crash, hundreds of gen-AI data policy violations, and Chinese cyberattacks against Taiwan intensified. The post In Other News: 8,000 Ransomware Attacks, China Hacked US Gov Emails, IDHS Breach Impacts 700k appeared first on SecurityWeek .
AWS Wickr versions prior to 6.62.13 on Windows, macOS, and Linux contain an improper resource release vulnerability in the call termination process. This flaw may allow a call participant to continue receiving audio input from another user after closing their call window under certain user actions. The issue affects AWS Wickr, Wickr Gov, and Wickr Enterprise desktop clients. Users are advised to upgrade to version 6.62.13 to mitigate this vulnerability.
Input from search query parameter in GOV CMS is not sanitized properly, leading to a Blind SQL injection vulnerability, which might be exploited by an unauthenticated remote attacker.
Versions 4.0 and above are not affected.
A new phishing campaign has been observed leveraging a .gov domain to deceive employees into believing they owe an unpaid toll. The scam uses urgency and fear tactics, threatening penalties or vehicle registration holds if the balance is not paid immediately. The threat actors utilize the GovDelivery system to increase legitimacy, despite using Indiana's instance for a Texas-related scam. The phishing link leads to a fake website that collects personal information and credit card details. The campaign exploits fear of consequences and mimics a well-known service, highlighting the importance of integrating human expertise into email security processes to identify threats that bypass conventional malicious indicators.