When the damage starts with a supplier or third-party software: liability, dependent business interruption and recourse.
A week dominated by LLM security: defenses for agents, for fine-tuning and for Android apps that integrate them, plus watermarking of generated text. On hardware, side channels on the NVLink interconnect between GPUs and a Spectre-v2 attack in JIT engines. A new class of XSS via DoH, mobile single sign-on defects and the verification of Signal's post-quantum ratchet round out the edition.
Research · Nov 15, 2026
ReadA large conference's proceedings inflate the counts, so the jumps should be read with caution. The content is strong on AI agents (hijacking Android GUI agents, defenses for LLM ones), on defeating memory protections (V8 sandbox, kernel MTE) and on side channels in cryptographic implementations and TEEs.
Research · Nov 15, 2026
ReadSince 12 November 2026 a server failure at Microsoft 365 has stopped Outlook, Exchange Online and other services in at least 74 countries, with no sign of an attack. The analysis: why many cyber policies do not trigger, waiting periods and quantification, aggregation on a single supplier and questions for renewals.
Insights · Nov 13, 2026
ReadOn 4 November 2026 Crédit Mutuel Arkéa disclosed that attackers who got in through a supplier's platform used the bank's own account to transfer more than 100 million euros, later converted into cryptocurrency. The analysis: loss of funds between crime and cyber policies, aggregation on the supplier, the speed of conversion, controls and questions for renewals.
Insights · Nov 11, 2026
ReadThis week three studies test AI agents: poisoned MCP tools that succeed in almost one case in four, signed tools that leak data and Office files the model reads differently from us. On hardware, a Trojan in sensors and two attacks on SLH-DSA and ML-KEM stand out. Compiler bugs in zkVMs and browser NativeMessaging round out the edition.
Research · Nov 08, 2026
ReadOn 29 October 2026 a vendor's monitoring script loaded on Polymarket was replaced and took about 2.7 million dollars from more than 700 accounts. The analysis: which policy responds, aggregation across third-party scripts, controls and the questions to bring to renewals.
Insights · Nov 04, 2026
ReadIn its letter of 7 July 2026, ECB Banking Supervision asked significant banks for an action plan against AI-enabled cyber threats, due by 31 October. The ESRB raised systemic cyber risk to "severe". An underwriter's reading: what the ECB asks for, what changes for cyber policies of financial institutions and which questions to bring to renewals.
Regulation · Nov 02, 2026
ReadFact sheet updated to 1 November 2026 on Regulation (EU) 2022/2554 (DORA) and Italian Legislative Decree 23/2025: who is in scope, ICT risk management, major incident reporting, critical ICT providers, fines and what it means for the cyber insurance of banks, insurers and intermediaries.
Regulation · Nov 02, 2026
ReadFact sheet updated to 1 November 2026 on the NIS2 Directive and Italian Legislative Decree 138/2024: who is in scope, security and incident reporting obligations, fines, management liability and what it means for cyber insurance.
Regulation · Nov 02, 2026
ReadFact sheet updated to 1 November 2026 on Regulation (EU) 2024/2847 (Cyber Resilience Act): which products are in scope, reporting of exploited vulnerabilities in force since 11 September 2026, security requirements from 11 December 2027, fines and what it means for cyber, tech E&O and product liability policies.
Regulation · Nov 02, 2026
ReadFrom 9 December 2026 software is a product under Directive (EU) 2024/2853. What falls into PI, product liability and cyber, where the three covers overlap or leave gaps, and what changes in underwriting questions.
Regulation · Oct 31, 2026
ReadAI agents and models are a new risk surface: a secret deleted from an agent's memory is recovered from summaries and fragments, a cloud provider infers sensitive data from a manipulated model's latency, AgentTriage reconstructs what an agent did. The second thread is social engineering. On post-quantum, a critical reading of the attacks on Classic McEliece.
Research · Oct 25, 2026
ReadOn 29 July 2026 an attack on CEVA Logistics stopped at least eight European warehouses, blocking goods of customers such as Bol, de Bijenkorf and Valve and exposing their customers' data. The analysis: goods stuck in someone else's warehouse, customer data at the supplier, aggregation in logistics and questions for renewals.
Insights · Aug 31, 2026
ReadOn 16 July 2026 Hugging Face disclosed that an autonomous AI agent, entering through a malicious dataset, had stolen credentials and internal data with thousands of actions in a few days. The analysis: the automated attacker, the AI supply chain, policy wording on AI and questions for renewals.
Insights · Jul 31, 2026
ReadBetween 27 May and 9 June 2026 ShinyHunters exploited an unknown Oracle PeopleSoft vulnerability against more than 100 organisations, two thirds of them universities, stealing personal data including sensitive data. The analysis: the MOVEit pattern for the third time, aggregation in education and questions for renewals.
Insights · Jun 30, 2026
ReadIn June 2026 coding agent extensions become a new supply chain dependency, and current detectors see only part of the risk. Partial ROV deployment enables BGP hijacks the victim cannot see, AirDrop and Quick Share show zero-click flaws, and a study reconstructs how law enforcement breached Encrochat.
Research · Jun 30, 2026
ReadBetween late April and early May 2026 ShinyHunters breached Instructure twice, maker of the Canvas platform used by thousands of schools and universities, and on 11 May the company paid to prevent publication of the data. The analysis: paying for extortion alone, one supplier for many insureds and questions for renewals.
Insights · May 31, 2026
ReadApril 2026 brings the IEEE S&P proceedings. Valid but inconsistent 5G messages crash the modems of 542 smartphone models, and imperceptible audio hijacks voice assistants. Computer-use agents cause harm even from benign instructions. On the post-quantum side, HQC and Falcon give way to side channels. The insurance lead market and the AI Act close the month.
Research · Apr 30, 2026
ReadOn 2 April 2026 CERT-EU attributed to TeamPCP the theft of data from the European Commission's cloud, entered with a key stolen through a compromised version of the open source tool Trivy: at least 30 EU bodies involved. The analysis: the security tool as the way in, shared infrastructure and questions for renewals.
Insights · Apr 30, 2026
ReadFebruary 2026 puts the AI agent supply chain centre stage: third-party skills, bundles of instructions and scripts, carry malicious code and injections that agents follow up to 80% of the time. Agents can exfiltrate data while their answer looks harmless, and LLMs make online deanonymisation a mass operation. Android updates and passkeys round out the month.
Research · Feb 28, 2026
ReadOn 6 February 2026 ransomware stopped the US payment platform BridgePay for at least four days: merchants forced to go cash-only, municipalities and water utilities unable to collect online. The analysis: the hidden supplier, waiting periods and deductibles, aggregation on payments and questions for renewals.
Insights · Feb 28, 2026
ReadOn 20 January 2026 an eScan antivirus update server distributed for two hours a malicious program that blocked updates and downloaded more code, hitting hundreds of computers in South Asia. The analysis: the trusted channel, aggregation on a security vendor, who pays and questions for renewals.
Insights · Jan 31, 2026
ReadJanuary 2026 is NDSS month. Malware brings its own signed, vulnerable drivers to switch off defences, and an architecture like ACE separates agent planning from untrusted apps. A handful of neurons hold up LLM alignment, desktop password managers give way to simulated keystrokes and email aliases confuse identities. IoT firmware, PLCs and PyPI packages round out the month.
Research · Jan 31, 2026
ReadOctober 2025 puts AI agents to the test: a benchmark measures twelve attacks through the Model Context Protocol, while quantization alone is enough to hide a backdoor in a model. Thirteen bytes send malware down the wrong path in Gmail's filter. The month also brings model signing on GPUs, LLMs against malicious npm updates and amplification through transparent DNS forwarders.
Research · Oct 31, 2025
ReadOn 15 October 2025 F5 disclosed that a state actor had stolen portions of BIG-IP source code and information on unpublished vulnerabilities, and CISA issued an emergency directive. The analysis: the theft that prepares the claims, state-actor exclusions, the loss to the vendor and questions for renewals.
Insights · Oct 31, 2025
ReadOn 1 September 2025 an attack stopped Jaguar Land Rover's factories for five weeks, with an estimated cost of 1.9 billion pounds to the UK economy and more than 5,000 businesses affected. The analysis: the production outage, the supplier chain, the cost of not being insured and questions for renewals.
Insights · Sep 30, 2025
ReadSeptember 2025 opens with EchoLeak, the first zero-click prompt injection exploited in a production enterprise assistant. The ACM CCS proceedings also bring a defence for computer-use agents, a systematic look at ten years of APT campaigns, attacks on SSH client signatures and on k-anonymised clinical data, and off-path TCP hijacking over shared IP addresses.
Research · Sep 30, 2025
ReadAugust 2025 questions some basic tools. CVSS, EPSS and other scoring systems rank the same vulnerabilities very differently, and AI model scanners miss most ways of poisoning them. Cloned-voice calls can be unmasked with listening challenges, and password strength meters reveal the passwords they were trained on.
Research · Aug 31, 2025
ReadBetween 8 and 18 August 2025 tokens stolen from Salesloft's Drift assistant gave access to the Salesforce data of more than 700 organisations, many of them security vendors. The analysis: OAuth authorisations, secrets in the CRM, aggregation, supplier liability and questions for renewals.
Insights · Aug 31, 2025
ReadJuly 2025 brings Rowhammer to the GPUs used for AI, with model accuracy dropping by up to 80%. Downloaded models can hide self-extracting malware, email-handling agents fall to a new kind of prompt injection, and an AI agent writes profitable smart contract exploits. On mobile, WebViews and SDKs bypass protections and over half of sensitive apps ignore secure hardware.
Research · Jul 31, 2025
ReadOn 3 July 2025 ransomware stopped ordering and licence provisioning at Ingram Micro, one of the largest technology distributors, for about six days. The analysis: dependent business interruption for resellers and managed service providers, waiting periods, aggregation in the IT channel and questions for renewals.
Insights · Jul 31, 2025
ReadJune 2025 shifts the focus to the physical world. Ship officers report spoofed GPS and ransomware on board, CAN bus data rebuild a car's route without GPS, and a video of a 3D printer is enough to steal its design. Computer-use agents fall for instructions hidden in screens, and fuzzing finds ten CVEs in 5G modems.
Research · Jun 30, 2025
ReadIn May 2025 Coinbase disclosed that support staff at an outsourced supplier, paid by criminals, had copied data on tens of thousands of customers, with a 20 million dollar demand refused and 307 million in costs. The analysis: dishonesty at the supplier, voluntary reimbursements, D&O and questions for renewals.
Insights · May 31, 2025
ReadIn April 2025 an attack that began with a password reset at an outsourced help desk stopped Marks & Spencer's online orders for 46 days, with an estimated impact of 300 million pounds and cyber cover of up to 100 million. The analysis: limit and loss, the help desk, aggregation by actor and questions for renewals.
Insights · Apr 30, 2025
ReadApril 2025 is the month of agents wired to external tools: MCP enables credential theft, a malicious document hijacks tool selection, and a new defence trains an LLM to spot prompt injection. Beyond AI, the XZ Utils backdoor is reconstructed step by step, AI-based weather forecasts can be faked, and post-quantum SSH gets its first full analysis.
Research · Apr 30, 2025
ReadOn 14 and 15 March 2025 the tj-actions/changed-files component, used by more than 23,000 repositories, was altered to print CI/CD secrets into logs; the first target was a Coinbase project. The analysis: costs without theft, the pipeline as a supplier, liability, aggregation and questions for renewals.
Insights · Mar 31, 2025
ReadOn 21 February 2025 Bybit lost about 1.5 billion dollars in ether during a transfer from its cold wallet, through a signing interface manipulated at the supplier Safe{Wallet}. The analysis: cyber, crime and custody policies, state attribution and questions for renewals.
Insights · Feb 28, 2025
ReadJanuary 2025 brings the NDSS proceedings. Open-model adapters can hide trojans, and the shared cache of LLM servers leaks other users' prompts. An autonomous car's LiDAR is spoofed at 60 km/h from 110 metres, attackers use Tor to hit cloudless IoT devices, and CSS alone is enough to fingerprint browsers and email clients.
Research · Jan 31, 2025
ReadIn January 2025 PowerSchool warned school districts of a data theft affecting tens of millions of students and teachers, which started from a support account. The analysis: supplier privileged access, ransom, notification, lawsuits and questions for renewals.
Insights · Jan 31, 2025
ReadIn December 2024 two vulnerabilities in Cleo's file transfer products were exploited at scale by Clop, with hundreds of claimed victims and confirmations such as Hertz and Chicago Public Schools. The analysis: data theft, insufficient patches, aggregation and questions for renewals.
Insights · Dec 31, 2024
ReadOn 21 November 2024 ransomware hit Blue Yonder's managed environment, stopping staff pay at Starbucks and fresh-food warehouses at Morrisons. The analysis: dependent business interruption, perishable stock, retail aggregation and questions for renewals.
Insights · Nov 30, 2024
ReadNovember 2024 shows LLM agents able to automate a penetration test and backdoors that leak documents from RAG systems. On the infrastructure side, a critical flaw in an RPKI validator may be an intentional backdoor. Nearly half the apps updating IoT devices have flaws, and an e-scooter becomes the target of battery ransomware.
Research · Nov 30, 2024
ReadOn 22 October 2024 Change Healthcare reported about 100 million notices for February's attack; the final count is 192.7 million. The analysis: concentration, dependent business interruption, notification, ransom and questions for renewals.
Insights · Oct 31, 2024
ReadSeptember 2024 marks AI security's shift from models to agents: web pages that steal personal data from agents and a defense based on information flow control. Social engineering gets automated with LLM-driven phone calls, while a field study scales back expectations for phishing training. In post-quantum, fault injection attacks hit ML-KEM and ML-DSA on microcontrollers.
Research · Sep 30, 2024
Read