Gemini CLI CVE, OpenAI Chip News, and AI Agent Security — 2026-06-24

Daily Tech Hack Global — public AI, technology, and cybersecurity news briefing for 2026-06-24. Global AI, technology, and cybersecurity briefing focused on AI developer tools, critical CVEs, AI infrastructure, and agentic operations. Central thesis: AI adoption is moving into CI, cloud operations, and infrastructure, so security depends on identity validation, narrow permissions, patched developer tools, and observable automation. Question answered: What changed today in AI and cybersecurity, and what should builders and businesses do now? Chapters: 00:00 Top three stories and why they matter 00:45 CVE-2026-12537: AI developer workflow risk 02:30 Other critical identity, cloud, and API CVEs 04:45 OpenAI chip and AI infrastructure trend 06:00 Agentic operations and observability 07:05 24-hour defensive checklist Sources and confirmed facts: 1. CVE-2026-12537 — Google Gemini CLI Publisher/source: NVD keyword GitHub URL: https://github.com/google-github-actions/run-gemini-cli/security/advisories/GHSA-wpqr-6v78-jr5g Confirmed facts: Improper Neutralization used in an OS Command in the container launcher in Google Gemini CLI (versions prior to 0.39.1) and run-gemini-cli GitHub Action (versions prior to 0.1.22) on headless CI platforms allows an unprivileged attacker to achieve pre-sandbox host-level code execution a maliciously crafted .gemini/.env file. Why it matters: This matters because modern AI and cloud workflows multiply the impact of identity mistakes, exposed secrets, unsafe package handling, and over-permissive automation. 2. CVE-2026-11718 — Google MCP Toolbox Publisher/source: NVD keyword identity URL: https://github.com/googleapis/mcp-toolbox/pull/3360 Confirmed facts: An authentication bypass vulnerability exists in the generic opaque token validation path (validateOpaqueToken) of googleapis/mcp-toolbox. When the toolbox validates an opaque token via an OAuth 2.0 introspection endpoint (RFC 7662), it decodes the response into an introspectResp struct. However, the subsequent claim-checking logic (validateClaims) evaluates the issuer condition as if a.issuer != "" && iss != "". If the external OAuth provider's introspection response omits the optional iss (issuer) field completely, the variable iss defaults to an empty string. This causes the conditional block to evaluate to false and be skipped silently. Consequently, the application accepts tokens issued by unauthorized or unintended third-party identity providers. Why it matters: This matters because modern AI and cloud workflows multiply the impact of identity mistakes, exposed secrets, unsafe package handling, and over-permissive automation. 3. CVE-2026-56270 — Flowise Publisher/source: NVD keyword GitHub URL: https://github.com/FlowiseAI/Flowise/security/advisories/GHSA-6pcv-j4jx-m4vx Confirmed facts: Flowise before 3.1.0 (versions 3.0.13 and earlier) contains a missing authentication vulnerability in the /api/v1/loginmethod endpoint that allows unauthenticated users to retrieve an organization's complete SSO configuration, including OAuth client secrets in cleartext, by providing an organizationId parameter. Remote attackers can send a GET request to harvest sensitive API credentials for Google, Microsoft/Azure, GitHub, and Auth0 integrations. This affects FlowiseAI Cloud and self-hosted instances where the endpoint is exposed. Why it matters: This matters because modern AI and cloud workflows multiply the impact of identity mistakes, exposed secrets, unsafe package handling, and over-permissive automation. 4. CVE-2026-27604 — FOSSBilling Publisher/source: NVD keyword identity URL: https://github.com/FOSSBilling/FOSSBilling/security/advisories/GHSA-57mv-jm88-66jc Confirmed facts: FOSSBilling is a free, open-source billing and client management system. Starting in version 0.5.4 and prior to version 0.8.0, an authorization bypass in the API role handling allows unauthenticated access to privileged `/api/system/*` endpoints. Because `system` resolves to the cron admin identity, attackers can invoke admin API methods without valid credentials, session, or CSRF token. Version 0.8.0 patches the issue. Some workarounds are available. Block external access to `/api/system/*` at reverse proxy/WAF, restrict API access by trusted source IPs only (`api.allowed_ips`), rotate all admin/client API tokens immediately, invalidate active sessions and reset high-privilege credentials, and/or review API request logs for suspicious `/api/system/` access and treat as potential incident. Why it matters: This matters because modern AI and cloud workflows multiply the impact of identity mistakes, exposed secrets, unsafe package handling, and over-permissive automation. 5. CVE-2026-48772 — ProxySQL Publisher/source: NVD keyword identity URL: https://github.com/sysown/pr

Видео Gemini CLI CVE, OpenAI Chip News, and AI Agent Security — 2026-06-24 канала dailytechhackglobal