Last updated: September 29, 2026 at 8:42 AM UTC
All 891 Vulnerability 357 Breach 144 Threat 383 Defense 7

AI evaluation startup Braintrust got hacked - and is asking every customer to rotate their AI provider API keys because the breached AWS account stored them all in one place

Braintrust, an AI evaluation and observability platform recently valued at $800 million, confirmed Tuesday that an unauthorized actor accessed one of its AWS accounts on May 4. The breached account held org-level API keys that customers store with Braintrust to access OpenAI, Anthropic, and other AI providers. Braintrust has confirmed exposure of one customer and is investigating three more reporting suspicious AI-provider usage spikes. The pattern - a relatively small AI infrastructure provider becoming a credential warehouse for downstream customers - is what Nudge Security's Jaime Blasco called 'the new shape of supply chain risk.'

Check
If your organization uses Braintrust, log into the org-level settings page and check the timestamp of every stored AI provider secret. Audit AI provider billing dashboards for unexpected usage spikes since April.
Affected
Braintrust customers, particularly AI-forward companies that store provider API keys in Braintrust org-level settings. Public reports suggest the customer base includes Box, Cloudflare, Dropbox, Notion, Ramp, and Stripe. Beyond Braintrust: any AI eval, observability, or gateway tool that holds customer-issued provider keys is the same risk pattern.
Fix
Rotate every AI provider API key stored with Braintrust - go to org-level settings, delete existing secrets, configure new ones, verify timestamps. Apply the same rotation to keys stored in similar AI eval/observability/gateway tools. Switch from static API keys to short-lived OIDC-issued credentials where the AI provider supports it. Add SCPs that restrict which AI provider services your IAM keys can call.

RansomHouse claims the Trellix breach and posts screenshots showing it reached internal VMware, Rubrik, and Dell EMC dashboards - far more than the 'small portion of source code' Trellix originally disclosed

Update on the Trellix breach we covered May 2: RansomHouse claimed the attack on its leak site Thursday and published screenshots that suggest the intrusion reached well beyond the source code repository Trellix originally acknowledged. Cybernews researchers reviewed the dumped images and identified internal dashboards for VMware vCenter, Rubrik backup, and Dell EMC storage - the systems that hold backups, credentials, and virtual machine images for the entire company. RansomHouse says the intrusion happened April 17 and resulted in data encryption. Trellix told BleepingComputer it's 'aware of claims of responsibility' and looking into them. RansomHouse currently lists 170+ victims on its Tor leak site.

Check
If your organization runs Trellix endpoint, IPS, ePolicy Orchestrator, or email security, audit checksums of every Trellix update installed since April 17. Hunt for unusual outbound traffic from Trellix product hosts.
Affected
Trellix customers - 53,000+ enterprises and government agencies in 185 countries protecting 200M+ endpoints. Acute risk: organizations relying on Trellix for backup integrity (Rubrik exposed) or VMware management (vCenter exposed). Defense and federal customers face higher residual risk pending Trellix's full incident report.
Fix
Hold non-emergency Trellix product updates until Trellix releases a written incident report with concrete scope. Verify checksums for every Trellix agent updated since April 17 against Trellix's published values. Treat any Trellix-issued credentials, API tokens, or signing certificates from before April 17 as potentially compromised and request rotation. Demand a written incident report within 30 days.

NVIDIA confirms a regional GeForce NOW partner in Armenia got breached - millions of user records exposed but NVIDIA's own systems are intact

NVIDIA confirmed Friday that a third-party GeForce NOW Alliance partner based in Armenia (GFN.am) was breached. The hacker, using the ShinyHunters handle on BreachForums, claims to have stolen names, email addresses, dates of birth, membership status, and 2FA enrollment status of millions of users - and is selling the database for $100,000. NVIDIA says its own systems are unaffected and the regional partner is notifying impacted users. The actor is suspected to be a ShinyHunters impersonator rather than the original gang. The partner serves users in Armenia, Georgia, Kazakhstan, Moldova, Ukraine, and Uzbekistan.

Check
If you or staff use GeForce NOW from Armenia, Georgia, Kazakhstan, Moldova, Ukraine, or Uzbekistan, log in to gfn.am and check for breach notifications. Search inbox for GeForce NOW or NVIDIA-themed emails since May 5.
Affected
GeForce NOW users registered through GFN.am, the Armenia-based regional partner serving Armenia, Georgia, Kazakhstan, Moldova, Ukraine, and Uzbekistan. Records reported leaked include names, email addresses, dates of birth, membership status, and 2FA enrollment - but not passwords. Acute risk for users who reused the GFN.am password elsewhere.
Fix
Reset GFN.am passwords and any other accounts using the same password. Enable 2FA if not already on. Treat any inbound emails referencing your real NVIDIA or GeForce NOW account details as hostile - go to gfn.am directly. For organizations: regional alliance partners often have weaker security than the parent vendor - audit which third-party regional services hold employee or customer data.

Hackers tell schools to pay by Tuesday or 275 million students' messages and IDs go public - Canvas operator Instructure confirms breach

Update on the Instructure breach we covered May 2: Instructure confirmed Saturday that names, email addresses, student ID numbers, and private messages between students and teachers were exposed. ShinyHunters now claims 275 million individuals across 9,000 schools worldwide are in the dataset, totaling 3.65+ TB of data including billions of private messages. The group set a pay-or-leak deadline of May 6 - this Tuesday. The Salesforce instance was also breached. This is Instructure's second breach in eight months. PowerSchool's January 2025 breach with similar scope produced a $17.25 million settlement.

Check
If your school or organization uses Canvas, prepare your student/parent breach notification template this week - Instructure data is likely to be public by Tuesday.
Affected
Schools, universities, and corporate training organizations using Canvas - 9,000 institutions globally, 275 million individuals. Acute risk for K-12 districts where data on under-13 students falls under COPPA and state student privacy laws (NY Education Law 2-d, California SOPIPA, ~130 similar state statutes). Salesforce-integrated Canvas tenants face additional exposure.
Fix
Rotate every Canvas API key and re-authorize integrations as Instructure has now mandated. Pull your district's Canvas data-sharing inventory and identify which downstream tools held copies. For K-12: prepare COPPA and state-AG notification templates now - PowerSchool's breach triggered class actions in 11 states. Brief students, parents, and faculty that any 'Canvas account verification' email this week is potentially hostile.

Commercial real estate broker Marcus & Millichap data leaked publicly - 1.8 million records including job titles for follow-on phishing

Marcus & Millichap customer data was leaked publicly after the company refused to pay ShinyHunters' extortion demand. Have I Been Pwned added the breach yesterday with 1,837,078 unique email addresses, plus names, phone numbers, employer names, job titles, and company addresses. Marcus & Millichap is a major US commercial real estate brokerage that closed $50.9 billion in transactions in 2025. The company says the leaked data 'appeared limited to company forms, templates, marketing materials, and general contact information' but ShinyHunters originally claimed 30 million Salesforce records. The leak extends the ShinyHunters wave that already published Pitney Bowes, Carnival, Udemy, ADT, and ZenBusiness.

Check
If you've ever interacted with Marcus & Millichap as a buyer, seller, or broker, watch for highly-targeted phishing referencing real property listings or transaction history over the next 90 days.
Affected
Marcus & Millichap clients - commercial real estate buyers, sellers, brokers, and prospects whose employer and job title data is now public. Acute risk: real estate scammers running 'wire transfer fraud' against named buyers using the leaked job titles and employer names to make spear-phishing convincing. Lenders and title companies that worked transactions with Marcus & Millichap face downstream exposure.
Fix
Treat any Marcus & Millichap email referencing your real role or company as potentially hostile - call known contacts via published phone numbers to verify. For real estate professionals: enable wire transfer verification protocols requiring out-of-band confirmation. Lenders and title companies should add Marcus & Millichap-themed lookalike domains to phishing detection. Affected individuals can monitor through HIBP.

Cybersecurity firm Trellix says attackers reached part of its source code repository

Trellix, the cybersecurity company formed from the 2022 merger of McAfee Enterprise and FireEye, disclosed Friday that attackers reached part of its source code repository. The company says it has 'no evidence' that source code releases were tampered with, that the source code itself was exploited, or that customer data was affected - but it has not said how long the attackers had access, who they were, or what they took. Trellix is now working with outside forensics firms and has notified law enforcement. Trellix sells endpoint protection, email security, and managed detection products to enterprise and government customers. The company has not given a timeline for further disclosure.

Check
If your organization uses any Trellix product, watch for unusual update patterns this week and avoid auto-updating until Trellix confirms the integrity of its release pipeline.
Affected
Trellix customers - enterprises and US government agencies that use Trellix endpoint, email, IPS, or managed detection products. Source code access doesn't automatically mean compromised products, but it's the starting position for finding new vulnerabilities. Defense and federal customers face higher residual risk pending Trellix's full disclosure.
Fix
Verify Trellix product update integrity by comparing checksums for any agent updated since the breach window. Hold non-emergency Trellix updates pending more clarity. For high-security environments, run Trellix in monitor-only mode for the next two weeks. Track Trellix's incident page directly and demand a written incident report within 30 days.

Mark Cuban-backed business filing service ZenBusiness leaked - 5 million customer records now public after ShinyHunters extortion failed

ZenBusiness customer data is now public on Have I Been Pwned, with 5,118,184 unique email addresses confirmed - alongside names, phone numbers, and CRM records pulled from Snowflake, Mixpanel, and Salesforce. ShinyHunters had threatened to publish the data in March after a failed extortion attempt; HIBP added the dataset yesterday. ZenBusiness is the AI-driven LLC formation and small business compliance platform backed by Mark Cuban. The breach extends the ShinyHunters wave that's already publicly released Pitney Bowes (8.2M), Carnival (7.5M), Udemy (1.4M), ADT (5.5M), and now ZenBusiness.

Check
If you used ZenBusiness to set up an LLC, treat any inbound communication referencing your real business name, formation date, or registered agent details as potentially hostile.
Affected
ZenBusiness customers - mostly small business owners, freelancers, and startup founders. The leak includes business formation details that uniquely identify the type of business you set up. Acute risk: small business owners targeted by 'compliance reminder' phishing referencing their real EIN, registered agent address, or annual report deadline.
Fix
Reset ZenBusiness account passwords and rotate any password reused on other accounts. Watch state filing systems for unauthorized changes to your registered agent or business address - attackers can hijack LLCs by changing these. Treat any 'urgent compliance notice' email as potentially hostile. For LLCs holding valuable assets, consider freezing changes through your secretary of state's office where supported.

Instructure, the company that runs Canvas for schools and universities, says hackers breached its systems

Instructure disclosed Friday that a 'criminal threat actor' breached its systems. The company runs Canvas, the learning management platform used by schools, universities, and corporate training programs - and a successful breach exposes student records, teacher records, course content, and grades. Instructure has not said how many users are affected or what data was taken, only that outside forensics are investigating. Canvas Data 2 and Canvas Beta have been in maintenance since May 1, with customers warned about API key issues. The pattern matches the January 2025 PowerSchool breach, which exposed data on 62 million students and is still being followed by ransom demands against individual schools.

Check
If your school or organization uses Canvas, audit which API keys you have integrated with Canvas and rotate any issued in the past 6 months as a precaution.
Affected
Schools, universities, and corporate training organizations using Canvas. Student records, teacher records, course content, gradebook data, and uploaded files are all in scope until Instructure confirms otherwise. Salesforce-integrated Canvas tenants may be at higher risk - 2025's Instructure incident traced to a Salesforce compromise.
Fix
Rotate Canvas API keys, especially for downstream tools (gradebook integrations, SSO, third-party plugins). Brief students, parents, and faculty that any 'Canvas account verification' email is potentially hostile - go to canvas.instructure.com directly. Request Instructure's incident notification timeline in writing and pre-prepare your own student/parent notification template.

France arrested a 15-year-old as the suspected hacker behind the French government ID agency breach - 11.7 million records confirmed stolen

Update on the ANTS breach we covered April 22: French police detained a 15-year-old on April 25, suspected of running the breach3d alias and stealing data from France Titres (ANTS), the agency that issues French ID cards, passports, and driver's licenses. The Paris Prosecutor's Office charged the minor on April 29 with three offenses carrying up to seven years in prison. ANTS now confirms 11.7 million accounts affected - lower than the original 19 million claim but still one of the largest leaks of French citizen identity data ever. Exposed data includes full names, email addresses, dates of birth, postal addresses, and phone numbers.

Check
If you live or operate in France, watch for highly-targeted phishing referencing real ID details over the next 90 days - the breach data is now confirmed in attacker hands.
Affected
11.7 million French residents whose ID data, contact details, and dates of birth are in the breach3d dataset. Acute risk for individuals who used these details to create accounts at French government services or banks. Organizations operating in France that use government-issued ID for KYC checks need to assume their data sources are tainted.
Fix
ANTS recommends affected users reset passwords on government and banking accounts and watch for impersonation messages claiming to be ANTS or La Poste. Treat any inbound email referencing your real ANTS data as hostile. KYC checks based on French government ID numbers should be backed by additional verification (face match, document liveness check) for the next 12 months.

Vimeo confirms user data was exposed via breach at analytics provider Anodot

Vimeo confirmed yesterday that user data was exposed when its analytics provider Anodot was breached. The video service hasn't said how many users are affected or what data was exposed beyond 'limited' account information, but Anodot's role suggests the leaked records include event-level user activity tied to Vimeo accounts: video views, account IDs, and the kind of telemetry analytics providers ingest. The pattern is the same as Citizens Bank, Frost Bank, Pitney Bowes, and now Vimeo: customer data leaks through a third-party vendor that the customer never directly signed up with.

Check
If you use Vimeo for any work-related video hosting, watch for Vimeo-themed phishing emails over the next few weeks referencing real account activity.
Affected
Vimeo users whose account data was processed by Anodot - a substantial subset given Anodot is a primary analytics provider. The risk is targeted phishing rather than account takeover: scammers who can reference real video views or account creation dates sound legitimate enough to bait credential resets. Organizations hosting marketing or training videos on Vimeo should expect staff targeting.
Fix
Treat any Vimeo email referencing your real account activity as potentially hostile - go to vimeo.com directly. Enable two-factor auth on Vimeo accounts, especially shared organizational ones. Review access logs for unfamiliar logins since April. For organizations: pull your vendor inventory and identify other analytics providers (Mixpanel, Heap, Amplitude) that hold customer data, and confirm breach notification SLAs.