Weekly Cybersecurity Intelligence Report: Cyber Threats and Breaches 10 Aug to 16 Aug 2026

The week of 10 to 16 August 2026 was defined by third-party access: every major incident reached its victims through a system the victims did not own. A cyberattack on Ceva Logistics exposed customer data across at least six downstream brands including Bol and Valve’s Steam hardware business. Trezor warned nearly 14,000 customers after its fulfilment partner ShipMonk was breached. A compromise of the Trivy scanner cascaded into malicious LiteLLM releases that CloudSEK says exposed more than 2,500 organizations. Wesco confirmed a cloud CRM incident after the ExfilSquad group claimed 2.6 million records, and the ransomware group behind the AnMed health system attack hijacked the hospital’s own Facebook page to press its demands. The shared mechanism is supplier and platform access, not perimeter compromise of the named victims themselves.

This weekly cybersecurity report covers confirmed incidents disclosed between 10 and 16 August 2026, with every figure traced to a primary source. Claims made only by attackers are labelled as claims.

1. Ceva Logistics breach exposes customer data at Bol, Steam, and four other brands across eight warehouses

Overview: On 11 August 2026, The Record reported that a cyberattack on French logistics giant Ceva Logistics disrupted eight European warehouses and exposed order data held for downstream retail customers, with Bol, De Bijenkorf, Ace & Tate, Ajax, Zalando, and Valve’s Steam hardware business notifying their customers. Ceva informed Bol of the compromise on 1 August 2026, and customer notifications rolled out during this reporting week.

Impact: Eight European warehouse operations were disrupted, and six named consumer brands issued notifications. Exposed data includes names, addresses, postal codes, phone numbers, email addresses, order numbers, tracking information, and purchase details such as hardware types and prices. Ceva retains customer order data for up to 90 days, which frames the exposure window. Neither Ceva nor its customers have published a total affected-customer count.

Details: Bol told customers that attackers gained access to two Ceva systems used to process orders. The initial entry point has not been disclosed, there is no public attribution, and it remains unconfirmed whether ransomware was deployed or an extortion demand was made. The disclosure stops short of explaining how long the attackers held access before the 1 August notification to Bol. This maps to supply chain compromise (MITRE ATT&CK T1195) from the perspective of the downstream brands.

Remediation guidance: Map which suppliers hold your customer data, what fields they hold, and for how long. Test supplier-facing integrations and data flows from the attacker’s perspective, not just your own perimeter. Set contractual breach-notification clocks measured in hours, since Ceva’s 1 August notice reached end customers only around 10 days later.

CISO takeaway: Your customers will hold you accountable for a logistics provider’s breach, and six brands spent this week proving it. Budget conversations about third-party and supply chain risk management should treat supplier systems that process your customer data as part of your own attack surface.

2. Malicious LiteLLM releases from the Trivy compromise expose more than 2,500 organizations

Overview: On 12 August 2026, SecurityWeek reported CloudSEK research showing that the earlier compromise of the Trivy vulnerability scanner cascaded into the LiteLLM open-source LLM gateway, pushing malicious versions 1.82.7 and 1.82.8 to PyPI. The packages carried information-stealing code that executed on every Python invocation.

Impact: CloudSEK counts more than 2,500 organizations impacted and 434,000 CI/CD pipelines exposed, with Nvidia, AWS, Samsung, Salesforce, and Cisco among the named exposures. The malicious packages were live for roughly 40 minutes before removal, which was enough for propagation at that scale. One attribution conflict is on the record: SOCRadar assesses that most of the 2,500 organizations were exposed earlier through the Trivy incident itself rather than through LiteLLM, and SecurityWeek published that clarification on 14 August 2026.

Details: The threat actor tracked as TeamPCP compromised Trivy through an unrevoked CI token, and the compromised scanner was then installed automatically inside LiteLLM’s build pipeline. CloudSEK summarised the chain as one unrevoked token, three tools deep. This is software supply chain compromise (MITRE ATT&CK T1195.002), and the AI tooling angle matters: the poisoned package sits in the dependency tree of production LLM applications.

Remediation guidance: Pin dependencies and verify hashes in every CI pipeline that builds AI or LLM tooling. Rotate and scope CI tokens now, since a single unrevoked token started this chain. Restrict outbound network access from build runners so an injected stealer has nowhere to send data.

CISO takeaway: A 40-minute window on PyPI reached 434,000 pipelines by CloudSEK’s count, which is the speed mismatch between package distribution and quarterly vendor reviews. The AI application stack your teams adopted this year inherits the supply chain risk of every scanner and gateway inside its build path.

3. Trezor warns nearly 14,000 customers after fulfilment partner ShipMonk breach

Overview: On 13 August 2026, hardware wallet maker Trezor disclosed that its fulfilment partner ShipMonk suffered unauthorized access exposing order and shipping data for purchases made between 10 May and 8 August 2026. Trezor stated that one of its shipping providers experienced a data breach that exposed sensitive order data, and confirmed its own systems and devices were unaffected.

Impact: Nearly 14,000 customers across seven countries are affected: 11,742 had names, email addresses, phone numbers, and shipping addresses exposed, and a further 1,947 had names, cities, and email addresses exposed. The affected countries are the United States, United Kingdom, Sweden, Colombia, Brazil, Italy, and Portugal. Trezor says this is the first exposure of customer phone numbers and shipping addresses in its 13-year history, and that customer funds and devices face no direct risk.

Details: The breach occurred at ShipMonk, not on Trezor infrastructure, so device firmware and recovery seeds are not in scope. The disclosure does not identify the intrusion method at ShipMonk, and no data has been confirmed published or sold as of 16 August 2026. For a crypto-hardware customer base, a name plus a shipping address is targeting data for phishing and physical coercion, which is why this incident matters beyond its record count.

Remediation guidance: Enforce data minimisation on fulfilment partners: they need a shipping label, not a permanent record. Set contractual retention limits so order data ages out in weeks, not quarters. Brief customer-facing teams on the phishing wave that follows any wallet-vendor breach, since attackers now hold verified buyer contact details.

CISO takeaway: The 90-day-style retention windows at logistics and fulfilment partners turn every one of your sales into a standing liability held on someone else’s systems. Purchase data tied to high-value asset ownership deserves the same vendor scrutiny as payment data.

4. Wesco confirms cloud CRM incident after ExfilSquad claims 2.6 million records

Overview: On 11 August 2026, industrial distributor Wesco International confirmed to BleepingComputer that it is investigating a cybersecurity incident affecting its cloud CRM environment, after the extortion group ExfilSquad claimed to have stolen 2.6 million records. Wesco stated it is aware of a claim of CRM data exfiltration by a third party.

Impact: The 2.6 million figure is the attacker’s claim and only the attacker’s claim; Wesco has not confirmed any record count. The company stated it does not believe payment card information, financial account information, or other sensitive customer or employee data is at risk, and found no ransomware or other malicious software on its systems. ExfilSquad published data on its leak portal after its ransom deadline expired.

Details: The claimed data set spans customer and employee PII, account and contact data, CRM user profiles, and authentication metadata. The intrusion method is unconfirmed. Researchers note ExfilSquad has previously harvested improperly configured Microsoft Power Pages data tables, which makes cloud application misconfiguration the leading hypothesis, but Wesco has not validated that vector. The gap between a 2.6 million record claim and a no-sensitive-data-at-risk statement is exactly what an investigation should resolve publicly.

Remediation guidance: Audit external sharing and anonymous-access settings on every SaaS CRM and low-code portal this month. Enumerate what an unauthenticated attacker can reach in your cloud CRM tenant, then test it. Prepare a disclosure position for extortion claims in advance, so the gang’s leak post is not your first draft.

CISO takeaway: Cloud CRM platforms are internet-facing applications and belong in your external attack surface management scope, not in a trusted-SaaS blind spot. Extortion groups increasingly monetise misconfiguration without deploying any malware, which defeats detection strategies built around payloads.

5. Ransomware group hijacks AnMed health system’s Facebook page as 11 of 106 facilities remain closed

Overview: On 11 August 2026, AnMed, a South Carolina nonprofit health system, identified unauthorized posts on its social media accounts after the ransomware group The Gentlemen hijacked its Facebook page to publish ransom demands, a follow-on to the 26 July 2026 malware attack that AnMed describes as a cybersecurity disruption. The hijacked account was subsequently removed.

Impact: At its 27 July peak the attack closed 83 of AnMed’s 106 facilities, or 78 percent. By 13 August 2026, 11 facilities remained closed and MyChart patient portal access had been restored with added security measures. The Gentlemen claim to hold 6 TB of data including HIV status, mental health, and sexual assault records; AnMed says those claims have not been verified and are under investigation. No patient data compromise is confirmed as of 16 August 2026.

Details: AnMed has confirmed malware but not the entry point or the strain’s technical chain. The Facebook hijack is the notable technique: seizing a victim’s own channels converts an extortion negotiation into a public pressure campaign aimed at patients. The Gentlemen claimed 125 attacks in Q2 2026, ranking among the most active groups of the quarter, so this playbook will repeat.

Remediation guidance: Put MFA and a hardware-key requirement on corporate social media accounts today, and pre-stage takedown contacts at each platform. Rehearse clinical continuity for a multi-week IT outage, since AnMed’s closure count stayed above zero for 18 days. Separate the credentials that manage public channels from the identity systems an intruder reaches first.

CISO takeaway: Extortion pressure now targets your audience, not just your data, and your social accounts are part of the incident surface. Recovery time is the number your board will remember: 83 facilities closed became 11 over roughly two weeks, and every day of that curve had a cost.