All news with #vulnerability disclosure tag

🔔 CISA added two vulnerabilities to its Known Exploited Vulnerabilities (KEV) Catalog after confirming active exploitation. The agency emphasizes these flaws are common attack vectors that present substantial risk to the federal enterprise. BOD 26-04 requires Federal Civilian Executive Branch agencies to prioritize rapid remediation of high-risk CVEs in the KEV catalog and to assess potential compromise before patching. CISA urges all organizations to adopt risk-based vulnerability management and to submit suspected exploited flaws via the KEV Nomination Form.

🔒 Researchers disclosed three patched vulnerabilities in LangGraph, including a critical SQL injection and unsafe deserialization chain that could enable remote code execution in self-hosted deployments. LangGraph is an open-source framework from LangChain for building stateful, multi-agent AI applications. Check Point and researcher Yarden Porat reported the issues, which affect SQLite and Redis checkpointers but not LangChain's managed LangSmith service.

🔒 ServiceNow notified customers after remediating a vulnerability that allowed an unauthenticated API endpoint to return tenant data under certain configurations. The issue, first reported via the vendor’s bug bounty program in April, prompted hosted updates on June 5 and guidance for self-hosted deployments. ServiceNow says affected instances were a subset of tenants and that observed activity appears linked to security researchers, though investigation continues. Customers are urged to apply updates and review logs for signs of unauthorized access.

🔔 CISA has added a new vulnerability to its Known Exploited Vulnerabilities (KEV) Catalog after observing active exploitation. The advisory reiterates that such vulnerabilities are frequent attack vectors and pose significant risks to the federal enterprise. It references BOD 26-04, which requires Federal Civilian Executive Branch agencies to prioritize rapid remediation of high-risk CVEs listed in the KEV catalog and to assess for compromise prior to patching. CISA urges all organizations to adopt risk-based vulnerability management and offers a KEV Nomination Form for reporting exploited vulnerabilities.

🛡️ Researchers describe a new "agentjacking" attack that tricks AI coding agents into executing arbitrary code by injecting malicious instructions into Sentry error events. Tenet Security says the flaw leverages Sentry DSNs — public, write-only credentials — to post crafted markdown that appears as legitimate remediation guidance. Agents retrieving unresolved errors via MCP render the injected content as trusted and may execute the embedded commands with developer privileges. The report confirmed high exploitability across popular agents and thousands of exposed DSNs.

🔒 Cybersecurity researchers disclosed six vulnerabilities in protobuf.js, the JavaScript/TypeScript implementation of Protocol Buffers, that can enable remote code execution (RCE) and denial-of-service (DoS) when untrusted schemas or payloads are processed. Named Proto6, the flaws affect Node.js apps, Google Cloud client libraries, messaging frameworks like Baileys, and CI/CD pipelines. Patches are available in protobufjs 7.5.6 and 8.0.2 and protobufjs-cli 1.2.1 and 2.0.2, and users are urged to update to mitigate risks stemming from trusting schema and metadata by default.

🛡️ A critical authentication bypass in phpBB forum software allows an attacker to hijack any account, including administrators, with a single unauthenticated request and no password. Tracked as PTT-2026-004 and rated 9.4, the flaw affects all versions up to 3.3.16 (and 4.0.0 alpha) using default database authentication, while a second OAuth-related issue (PTT-2026-005, 8.3) can bind attacker credentials via CSRF and missing state checks. phpBB released 3.3.17 on June 6 to fix both issues and urged immediate upgrades; temporary mitigations include disabling OAuth and auditing OAuth bindings.

🔒 On May 29, researcher Taylor Hornby discovered a critical flaw in Zcash's Orchard shielded pool; the Zcash team had contracted him specifically for this audit. The vulnerability involved a missing enforcement in a validation check that could have allowed creation of ZEC out of thin air despite valid-looking zero-knowledge proofs. The issue has been patched, but there is no reliable way to determine whether the bug was exploited prior to the fix.

🔒 Check Point has reported active exploitation of a critical logic flaw in certificate validation affecting Remote Access and Mobile Access VPNs configured to use deprecated IKEv1. The issue, tracked as CVE-2026-50751 (CVSS 9.3), lets unauthenticated attackers bypass user authentication and establish VPN sessions without valid passwords. Exploitation requires IKEv1 enabled, legacy clients accepted, and no machine certificate requirement; activity was first observed in early May 2026 and has targeted a few dozen organizations globally.

🛡️ Researchers at Cyera disclosed six vulnerabilities in protobuf.js, a JavaScript implementation of Google’s Protocol Buffers, that allow untrusted schema data to influence application behavior. The most severe issues enable code generation and injection via manipulated schema metadata, potentially leading to remote code execution when crafted inputs are accepted. The flaws affect protobuf.js versions up to 7.5.5 and 8.0.1 and also impact protobuf.js-cli; patches are available in updated releases.

🔒 Meta disclosed that a vulnerability in its AI-assisted High Touch Support (HTS) tool allowed threat actors to reset passwords and hijack over 20,000 Instagram accounts. Attackers exploited HTS by submitting email addresses not verified against target accounts, obtaining reset links for accounts without 2FA. Meta disabled the HTS system, invalidated generated reset links, secured impacted accounts, and required affected users to reset passwords and re-authenticate. The company said it will fix the verification check and review similar recovery flows across its platforms.

🛡️ depthfirst's autonomous agent discovered 21 previously unknown zero-day vulnerabilities in FFmpeg, producing reproducible PoC inputs for each at a reported cost of about $1,000 for the run. In the same week, Google released Chrome 149 with fixes for a record 429 security bugs, over 100 of which are critical or high severity, following an overhaul of its bounty program to cope with a surge of AI-generated reports. The findings illustrate how AI is accelerating vulnerability discovery and increasing pressure on triage and patching processes across widely used software.

🔒 A researcher discovered a vulnerability in Anthropic's Claude Code GitHub Action that allowed takeover of public repositories via a single opened GitHub issue. Anthropic patched the core bypass in January and released fixes in claude-code-action v1.0.94, rating the issue 7.8 under CVSS v4.0 and issuing a bounty. The flaw arose from overly permissive triggers that trusted actors ending in "[bot]" and example workflows allowing non-write users, enabling indirect prompt injection to exfiltrate environment secrets and OIDC credentials. Administrators should update to v1.0.94, audit workflows for untrusted inputs, and remove unnecessary permissions and tools to prevent exfiltration.

🛡️ A high-severity RCE vulnerability in Hugging Face Transformers lets attackers leverage a hidden config field to execute remote code when loading models. The flaw abuses an underscore-prefixed parameter, _attn_implementation_internal, bypassing trust_remote_code=false and triggering unsandboxed kernel downloads. A silent patch was released in Transformers 5.3.0; users should upgrade and scan cached configs.

🔒 NAVTOR NavBox through version 4.16.1.20 contained hard-coded credentials in its Windows Communication Foundation (SOAP) implementation. If SOAP is enabled, a local attacker could extract those credentials to authenticate to privileged WCF methods and write or overwrite files within application-defined paths, disrupting operations. NAVTOR released a patch in April 2026; versions 4.17.2.6 and later include the fix, and connected NavBox users will be updated automatically.

🛡️ A newly discovered DoS technique called HTTP/2 Bomb can bring down default HTTP/2 deployments of major servers (NGINX, Apache, IIS, Envoy, Cloudflare Pingora) from a single machine in seconds. Discovered with assistance from OpenAI's Codex and reported by Calif researchers, it combines HPACK compression amplification with flow-control stalling to force massive memory allocations and prevent their release. Proof-of-concept exploits exist and patches or mitigations are partially available.

🔒 A two-year use-after-free vulnerability (CVE-2026-23479) in Redis 7.2.0 through older stable branches enabled authenticated remote code execution and was disclosed after Team Xint Code and Wiz published a full technical write-up. The flaw arises in unblockClientOnKey() where a freed client structure is reused, and the exploit chain leaks a heap pointer via Lua, forges a client, and corrupts memory accounting to overwrite a GOT entry. Default deployments and many cloud instances lacking passwords increase exposure; Redis released fixes on May 5 and recommends minor upgrades per series and tightened ACLs if immediate patching is not possible.

🛡️ Acer confirmed it's addressing two maximum-severity zero-day vulnerabilities in Wave 7 mesh routers. Reported by researcher Gergo Pap, both affect firmware T7c_GBL_1.01.000055 or earlier and permit remote, unauthenticated access to sensitive data and persistent backdoors. Acer plans firmware fixes by the end of June 2026 and urges users to update once patches are released and to disable or restrict remote management as a temporary mitigation.

🔒 Researchers disclosed an unpatched Windows issue that can expose a user's NTLMv2 hash via the search: URI handler. Similar to CVE-2026-33829 in the Snipping Tool, the flaw leverages a crumb=location: parameter to force an SMB connection and trigger NTLM authentication. The weakness produces the same Net-NTLMv2 leak and attack prerequisites, and Microsoft declined to patch it after responsible disclosure.

🛡️ Anthropic has broadened Project Glasswing, giving 150 additional organizations access to its most capable model, Claude Mythos Preview, to help find vulnerabilities in critical software. The program, first opened to roughly 50 partners in April, claims more than 10,000 high- or critical-severity flaws discovered to date. New participants span 15+ countries and underrepresented sectors like power, water, healthcare and hardware, chosen for the potential catastrophic impact of breaches. Anthropic warned that while discovery is accelerating, safe public release of Mythos-class models remains restricted due to incomplete safeguards.