What to look out for today
Today’s theme is speed: attackers are using stolen credentials and API tokens to move fast (minutes, not days), plus there’s a reminder that supplier mistakes (like leaked keys or tampered packages) can quickly ripple into customer environments.
- Stolen tokens and MFA phishing: attackers grabbing GitHub/API tokens or tricking staff into approving sign-ins can lead to rapid access and data theft.
- Software supply-chain incident (dev/AI tooling): a leaked master key was reportedly used to push a malicious package release in the LiteLLM ecosystem—relevant to anyone using AI/dev libraries through automated dependency updates.
- Actively exploited flaws flagged by CISA: reports of exploitation affecting Collabora Online Built‑in CODE Server and Arista VeloCloud Orchestrator (important if you or your IT provider run these).
- Google AI features and data sharing choices: if staff enable personalised AI features in Search/Gemini with connected services (e.g., Gmail/Photos), it may affect what data is shared and how ads are targeted—this is a governance and privacy decision for organisations.
Why this matters to smaller businesses
SMEs often rely on SaaS, outsourced IT, and “auto-updating” software stacks. If a single account, token, or supplier component is compromised, attackers can:
- access email, files, CRM or code repositories and quietly export data;
- send convincing internal/phishing emails from real accounts (leading to invoice fraud);
- tamper with software builds or updates that then spread the impact further.
Because these attacks can unfold quickly, the biggest gap is often time-to-detect and time-to-disable access.
Warning signs
- Unexpected MFA prompts or staff reporting “MFA fatigue” (many prompts in a short period).
- New logins from unusual locations/at unusual times, or “impossible travel” alerts.
- New API keys/tokens created, permissions widened, or security settings changed without a ticket/change request.
- Unplanned dependency/library updates in projects, or builds suddenly pulling in new versions without explanation.
- Users enabling new AI/personalisation features without business approval, especially if it requests access to other services’ data.
How attackers may exploit the situation
- Token theft: stealing GitHub, cloud, or SaaS tokens (from laptops, logs, chat tools, browser storage, or poorly stored secrets) to bypass password resets.
- MFA phishing: luring a user to a fake sign-in page and capturing session details or pushing repeated MFA prompts until one is approved.
- Supply-chain tampering: getting malicious code into a package/update so downstream organisations ingest it during normal updates or CI/CD builds.
- Exploitation of internet-facing management tools: if a supplier product is exposed online (often via an MSP), exploitation can become an entry point into multiple customer environments.
What to do today
- Run a quick access review: list who has admin access to email, file storage, finance platforms, and code repos; remove stale accounts and enforce least privilege.
- Rotate and re-check high-risk tokens: prioritise GitHub, CI/CD, cloud admin, backup system, and any “master keys” used by automation.
- Turn on/verify alerting: alerts for new OAuth app grants, new API keys, MFA method changes, and unusual sign-ins.
- Dependency hygiene: pause “auto-merge” dependency updates where possible; require review for new/unknown packages; confirm you can identify what version was deployed and when.
- AI feature governance: decide whether staff can connect business accounts (e.g., Gmail/Workspace) to personalised AI features; document an approved approach.
- Prepare comms: remind staff: don’t approve unexpected MFA prompts; report them immediately.
Ask your IT provider
- Can you show us recent sign-in alerts and how quickly you respond to suspicious logins?
- Do we have conditional access (location/device rules) and is it enforced for admins?
- How do you manage and rotate API tokens/service accounts used by integrations and automation?
- Do we run or use Collabora Online (Built-in CODE Server) or Arista VeloCloud Orchestrator anywhere (directly or via an MSP), and are they internet-exposed?
- What controls do we have to reduce supply-chain risk (dependency pinning, approval gates, integrity checks, and rollback capability)?
Patch watch - only one short paragraph, and only if relevant
CISA has flagged active exploitation affecting Collabora Online Built‑in CODE Server and Arista VeloCloud Orchestrator. If you use these products (often via an MSP), treat this as urgent: confirm exposure, apply vendor fixes/mitigations, and review logs for unusual admin activity.
One action today
Ask your IT provider today for a 15‑minute review of admin accounts and active API tokens (email, file sharing, finance, code/CI), and rotate anything high-risk or unknown.
Related Actions On Cyber resource
Actions On Cyber checklist: “Rapid account takeover response” (MFA prompt handling, token/key rotation, and emergency access review)
Sources
- Machine speed attacks require machine speed detection and response (BleepingComputer)
- CISA Flags Two Actively Exploited Critical Flaws in Collabora Online and Arista VeloCloud (The Hacker News)
- Leaked LiteLLM Master Key Let Attackers Steal GitHub Tokens and Hijack Release Pipeline (The Hacker News)
- Google Says Gmail and Photos Data Shared With AI Search Could Be Used for Targeted Ads (The Hacker News)
This brief is for general awareness and does not replace advice from your IT provider, legal adviser, insurer or incident response specialist.