the daily brief

Cyber / Brief, 2 Nov 2026: FBI blames North Korea for the $6 billion Binance theft

curated by Paolo De Rosa updated 02 Nov 2026

The FBI has pinned the theft of more than $6 billion from Binance, the largest crypto heist on record, on North Korea's TraderTraitor, publishing the wallets it wants firms to block, while Anthropic's own figures show AI now leading 26% of its AI research work, up from 1% in March, and a co-founder expects self-improving systems in 2027. In Europe the fight over who controls data and technology sharpened: the EU's data protection supervisor attacked the planned deal giving US authorities access to national police databases, Hungary signed a pact with Washington to pursue left-wing groups, and a draft Quantum Act abandons regulation for funding and buy-European procurement. A trojanised Claude Code update reached up to 5,000 developers through Cursor's marketplace, ESET documented the first destructive attack in Europe by a Belarus-aligned group, and a Cleveland hospital system was still rescheduling patients' procedures a week after a cyberattack. Britain, for its part, wrote 'control and sovereignty' into its AI principles, as the Pentagon quietly asked Baltic allies to tone down their warnings about Russia.

AI & Power

The Smartest People in AI Believe This Is the World’s Most Important Chart — Technology - WSJ.com
Why it matters: Anthropic's own data show AI leading 26% of its AI research work in August, up from 1% in March and none a year ago, though none is yet fully autonomous; co-founder Jack Clark now expects recursive self-improvement in 2027.
The figures come from Claude agents that classified 15,000 internal R&D tasks on a six-level scale, so the measure is the company's own and its method is AI-run, The Wall Street Journal reports. 'AI leads' means the model does most of the work under human supervision. OpenAI says it sees early signs too. Anthropic warns that this could come sooner than most institutions are prepared for, and no EU rule yet treats self-improvement as a reportable capability; the AI Act's systemic-risk regime is the nearest hook.

‘Control and sovereignty’ to shape UK approach to AI — myFT following
Why it matters: Andy Burnham's cabinet has approved seven principles for UK AI, including 'control and sovereignty matter' and infrastructure that must 'serve a national purpose', after a split between the culture and business departments.
Culture Secretary Lisa Nandy argued for a slower approach after creative industries objected to scraping, while the business department pushed for speed, the Financial Times reports. The Budget gave £700 million to the AI Security Institute, a biosecurity lab and a national cyber shield, plus a data-centre 'covenant' trading grid priority for community and environmental duties. Questions on sovereign capacity remain open. London now uses Europe's vocabulary of sovereignty and will make AI the focus of its G20 presidency next year.

Trump Administration Refuses to Release Information on AI Testing Framework — The Information
Why it matters: The Trump administration declined to produce key documents on its voluntary AI testing framework, despite a September agreement with the Protect Democracy Project after a freedom-of-information lawsuit.
A letter sent to the nonprofit on Friday refuses the records, The Information reports. The framework is the federal government's main window on frontier model safety, and it rests on voluntary commitments. Allies relying on US evaluations, the EU's AI Office among them, cannot see how it works. Excerpt-based.

Behind the Curtain: OpenAI's post-election power play — Axios
Why it matters: OpenAI plans a policy and PR campaign right after the US midterms, including rebranding data centres as 'AI compute campuses', to get ahead of a likely Democratic House and the 2028 field.
Global affairs chief Chris Lehane calls Tuesday an 'electoral portal' from one era to the next, and the message will be that AI is being built for people, Axios reports; Meta is also strengthening its ties with Democrats. It shows how much the labs expect the political climate on AI to harden. European regulators should expect the same framing in Brussels.

Making it easier for websites to express AI training preferences — OpenAI News
Why it matters: OpenAI says it will honour a shared robots.txt signal, developed in the IETF's AI Preferences group, that lets sites state one training preference for all participating AI providers.
The signal covers foundational generative AI training, will be implemented in the coming months and adds to OpenAI's existing crawler controls, the company says; it is voluntary. A machine-readable opt-out is what the EU's copyright rules and the General-Purpose AI Code of Practice ask providers to respect, so a common standard would matter most for European publishers.

Was I wrong about open models & cyber risks? — Interconnects AI
Why it matters: Nathan Lambert, a leading advocate of open models, reassesses his view of their cyber risk after 'Black October', the month of AI-driven vulnerability disclosures.
He argues that defenders are working harder behind the scenes and that the rise in flaws will take a long time to level off, Interconnects reports. His rethink matters because the debate decides how far open-weight models, Europe's route to AI sovereignty, can be released without restriction.

Organisations need to prove that their AI agents can shut down — The Strategist
Why it matters: ASPI argues that no AI agent in national security or critical infrastructure should get system access until a 'stand-down drill' proves that it, and agents working with it, will actually stop.
The test checks that an agent told to stop no longer has access and is no longer acting, including when agents work as a team, The Strategist writes. It is a concrete control that NIS2 entities could adopt now; the AI Act asks for human oversight but does not prescribe such a drill.

EU & Technology

Forget AI, Europe has a plan to win the next tech revolution — Technology – POLITICO
Why it matters: A draft of the Commission's Quantum Act, due in November, drops the AI Act's regulatory approach for funding and scale, and would oblige governments to buy from EU firms or countries with a strategic quantum partnership.
'It's not about regulation. There's nothing to regulate', a senior Commission official told POLITICO; the draft favours EU-level support so national champions can scale. Executives at France's Alice & Bob and Finland's IQM say Europe can lead. The procurement clause is less absolute than the leak we reported on 30 October, since it also admits strategic partners. Who counts as a partner will decide whether US firms are in.

„Grenzpartnerschaft“: Auch der EU-Datenschützer rügt geplantes Polizeiabkommen mit den USA — netzpolitik.org
Why it matters: The European Data Protection Supervisor criticises the planned EU–US framework giving American authorities read access to national police databases, warning of mass transfers of biometric data.
Washington has made the 'Enhanced Border Security Partnership' a condition of the Visa Waiver Program, with a deadline at the end of 2026, netzpolitik.org reports; the Commission published a signature-ready text on 10 September. It would apply provisionally on signature, with Parliament's consent afterwards. Wojciech Wiewiórowski wants 'public order' read narrowly, since searches are allowed on a broad threat test. The final draft bars queries by face image but lets replies include biometric photos; Berlin keeps its talks secret even from parliament.

Datentausch und Strafverfolgung: Trump findet ersten EU-Partner zur Jagd auf Antifa-Gruppen — netzpolitik.org
Why it matters: Hungary signs a memorandum with the US to share information on and jointly pursue 'left-wing extremist' networks, the first known bilateral deal of its kind and aimed at groups including Germany's 'Antifa Ost'.
Foreign ministers Anita Orbán and Marco Rubio signed it in Washington on 29 October, netzpolitik.org reports, under the US campaign that follows the NSPM-7 memorandum; Germany is preparing a second summit with the US on left-wing extremism. A member state agreeing to share data on political groups with a US programme criticised as surveillance of dissent raises questions under the GDPR and the Law Enforcement Directive.

EU and Canada’s 6-month test of commitment — Technology – POLITICO
Why it matters: The Montreal summit gives the EU and Canada six months to define the closest possible relationship, with Carney putting a deadline on 'associate membership' and a framework for EU investment in Canadian critical minerals.
Both sides want to shore up supply chains and cut reliance on the US and China, POLITICO reports; the European Investment Bank will fund minerals projects and a study will look at LNG exports. Canada is also joining EuroHPC. The test is whether 27 governments can agree on what the label means.

How Ireland plans to beat the data center backlash — Technology – POLITICO
Why it matters: Ireland, which halted new data-centre grid connections around Dublin in 2021, now makes tech firms fund new renewables as a condition of connection, an experiment other countries are watching.
The effective moratorium was lifted last year in favour of a checklist of duties, POLITICO reports, as pushback spreads across Europe and becomes a US midterm issue. It is an alternative to the Commission's push to build AI gigafactories quickly, and a model for grid-constrained member states.

Merz ‘surprised’ by Asian pushback to ‘Made in Europe’ plan for car industry — Policy – POLITICO
Why it matters: Friedrich Merz says South Korea and Japan protested strongly against the Franco-German plan to limit EU car subsidies to vehicles made in the EU, Switzerland and the UK.
South Korea's president asked to be treated like a member state, POLITICO reports. 'Made in Europe' clauses are spreading from cars to quantum and defence, and partners that Brussels courts on economic security are the ones objecting.

Philip R. Lane: AI adoption in Europe — ECB - European Central Bank
Why it matters: ECB chief economist Philip Lane says AI adoption, not frontier development, will decide Europe's productivity, noting that euro area digital investment rose 73% in real terms since 2014 to 3.3% of GDP, but much more slowly than in the US.
European firms are part of the AI supply chain and European investors help fund the global build-out through equity, debt and banks, so the euro area is also exposed to its financial risks, the speech says. It complements ESMA's warning on AI concentration.

US & Technology

DOJ raises privacy concerns over potential sale of Character.AI chats — The Record from Recorded Future News
Why it matters: The US Justice Department's bankruptcy office asks for a privacy ombudsman as Character.AI's user chats, often sensitive and sometimes with minors, may be sold in its bankruptcy under an 'ambiguous' privacy policy.
The company has 12 million monthly users, and the judge could block the sale of the data, The Record reports; state attorneys general have raised the same concern. Chat histories as a bankruptcy asset are a new privacy question; under the GDPR, a sale for a new purpose would need its own legal basis.

California officials confirm grand jury subpoenas as states, civil groups brace to protect elections from federal encroachment — CyberScoop
Why it matters: Federal grand jury subpoenas reached California election offices days before the vote, tied to the Justice Department's demand that states check voters against federal lists.
Los Angeles County's registrar calls it an attempt to intimidate officials into accepting federal supremacy over voter registration, CyberScoop reports. Voter-roll data is the focus. A federal judge separately blocked threats of prosecution against election workers.

What Democrats Can Learn From New York’s Data Center Moratorium — WIRED
Why it matters: Four months into New York's one-year data-centre moratorium, the first of any state, campaigners fear Governor Hochul will not sign a stricter bill, as anti-data-centre candidates look set to win on Tuesday.
WIRED reports doubts over how the state evaluates projects. Data centres are becoming a voting issue in the US as they are in Ireland and the UK, which bears on the EU's own compute plans.

China & Technology

Apple confirms Qwen integration for China AI rollout, with beta testing starting November 4 — TechNode
Why it matters: Apple will start testing Apple Intelligence in mainland China on 4 November with Alibaba's Qwen built into Siri, Writing Tools and Visual Intelligence, ending a two-year wait, with wider release in December.
Users can sign in with a Qwen account or with Apple, TechNode reports. An American platform now ships a Chinese model to meet Chinese rules, the mirror image of the US ban on DeepSeek, and shows how AI stacks split by jurisdiction.

Self-improving AI the key step towards AGI but safety is key: Z.ai founder — Tech - South China Morning Post
Why it matters: Z.ai co-founder Tang Jie says recursive self-improvement is the main research problem for reaching AGI and must be done safely, putting a leading Chinese lab on the same path as OpenAI and Anthropic.
He spoke in Shanghai on Saturday, SCMP reports; Cambricon's co-founder separately called for shared ethics standards with US rivals. Chinese labs are stating the goal openly as the US–China incident channel remains empty. Excerpt-based.

DeepSeek unveils AI ‘super agent’ in coding race with OpenAI, Anthropic — Tech - South China Morning Post
Why it matters: DeepSeek previews a 'super agent' that coordinates several coding agents with minimal human supervision, arguing that people can oversee only three to five agent sessions at a time.
The team has open-sourced a research prototype, SCMP and TechNode report. Removing the human bottleneck is the stated design goal, which sits awkwardly with the human oversight that the AI Act assumes. Excerpt-based.

Catch Us If You Can — The Wire China
Why it matters: The trial in United States v Huawei Technologies is under way in a federal court in Brooklyn and is shedding light on the company's practices, The Wire China reports.
The case, on fraud and sanctions charges, could put evidence about Huawei's corporate structure on the public record, relevant to European decisions on high-risk vendors. Excerpt-based.

U.S. Export Controls Fuel the Rise of Asian Data Centers Serving Chinese AI — The Wire China
Why it matters: Chinese AI firms are the top customers of Singapore-based DayOne's data centres, The Wire China reports, as US controls push Chinese compute offshore; DayOne hopes for US support.
Offshore compute is the gap in chip controls that the 'remote access' debate is about, and it runs through jurisdictions allied with the West. Excerpt-based.

La Liga goes dark in China over Mbappé comment — Semafor
Why it matters: China's state broadcaster dropped La Liga matches on Sunday, widening the row over Kylian Mbappé's joke about Mao from one player to a whole Spanish competition.
The escalation came after the apologies, Semafor reports. Beijing's commercial pressure on a European league continues even after compliance, which is the lesson European firms will draw.

Defence & National Security

Pentagon tells Europeans to cool their rhetoric on Russia, sources say — Defense News
Why it matters: Pentagon officials have asked several NATO allies, including Lithuania, Latvia and Estonia, to tone down public warnings about the Russian threat, four sources told Reuters.
The messages come as Washington restarts its push for a negotiated end to the war, mixing sanctions relief for Russia with pressure on Ukraine, Defense News reports. Asking frontline allies to soften threat assessments cuts against the case for European rearmament they are making at home.

British spy agency cannot account for thousands of data storage devices, watchdog says — The Record from Recorded Future News
Why it matters: GCHQ cannot account for thousands of USB sticks, hard drives and other storage devices, some possibly holding intercepted data, and the UK's surveillance watchdog says it is 'very concerned' at the slow investigation.
The Investigatory Powers Commissioner said the agency has no clear view of the scale, location or recoverability of the devices, and that the issue was raised years ago, The Record reports. A partner of many European services has lost track of the media it uses.

New UK policy requires foreign defense contractors to invest locally for all £50M+ deals — Breaking Defense
Why it matters: From April 2027 foreign firms winning UK defence contracts of £50 million or more must invest locally, up to the value of the contract, through factories, suppliers and technology transfer.
The Ministry of Defence says it continues to prioritise buying British, Breaking Defense reports. London adds its own offsets rule as it argues against 'Made in Europe' limits in EU defence funding.

Analysts find looming launch crisis is "more severe" than anticipated — Ars Technica - All content
Why it matters: Quilty Space estimates a shortfall of about 150 launches outside China through 2030, possibly double that with US military demand, and sees no end to the crisis this decade.
There is no evidence Starship will launch traditional satellites, Ars Technica reports. Europe's plans for IRIS² and sovereign constellations depend on launch capacity it does not control.

Ukraine completes work on new missile defense system, awaits European partner components — Defense News
Why it matters: Ukraine says it has finished its part of Freyja, the ten-country anti-ballistic project, and is waiting for European components from firms such as Kongsberg and Diehl before tests due by year end.
Fire Point builds the FP-7.X interceptor as Patriot stocks run short, Defense News reports. The schedule is the partners' claim and the system is untested; it would reduce Europe's reliance on US interceptors if it works.

Digital Sovereignty & Identity

UK facial recognition rollout widens as police oversight rules take shape — Biometric Update
Why it matters: Suffolk police will use live facial recognition in three towns this week, after August operations that scanned 82,590 faces and led to four arrests, two of them with no recognition alert.
The vans and NEC equipment are on loan from Bedfordshire Police, Biometric Update reports, as UK oversight rules take shape. The hit rate shows the scale of scanning behind each arrest. The AI Act restricts real-time biometric identification in public that the UK now treats as routine.

US Treasury makes ID.me mandatory as federal policy shifts toward Login.gov — Biometric Update
Why it matters: Since 28 October the private ID.me service is the only way to log in to TreasuryDirect, drawing questions from senators Wyden and Warren as federal policy moves towards the government's own Login.gov.
They ask Treasury to postpone until after the 2027 tax deadline and want a reply by 9 November, Biometric Update reports. A state tying access to savings to a private identity firm is the dependency the EU's public EUDI Wallet is meant to avoid.

ETSI releases specification to help mitigate AI deepfake risk — Biometric Update
Why it matters: ETSI publishes TS 104 275, a specification for reducing the risk that generative AI produces harmful deepfakes, treating mitigation as a duty across the AI system's lifecycle, not only detection afterwards.
It builds on ETSI's EN 304 223 baseline for AI security and supports objectives of the AI Act, Biometric Update reports. A European standard body setting duties for developers gives providers a route to show compliance with the Act's transparency rules.

Brazil mandates SIM registration with face biometrics, Level 2 PAD — Biometric Update
Why it matters: Brazil's telecom regulator Anatel will require face biometric checks for mobile phone registration from next year, with presentation-attack detection at level 2 or higher.
Operators can use a government database or a private provider, Biometric Update reports. Mandatory biometric SIM registration in a large democracy is a precedent that privacy groups will cite in EU debates on age and identity checks.

Türkiye plans digital ID wallet pilot for 2027 as age verification expands — Biometric Update
Why it matters: Türkiye plans a national digital identity wallet prototype in 2027, holding ID documents and diplomas, led by its Cybersecurity Presidency and starting with public employees.
The plan appears in the 2027 Presidential Annual Program, Biometric Update reports, alongside wider age verification. A wallet run by a security body differs sharply in governance from the EUDI model, even if the formats converge.

Threat Intelligence (CTI)

[P1] FBI Links North Korea's TraderTraitor to Record $6.43 Billion Binance Crypto Heist — The Hacker News
Why it matters: The FBI attributes the Binance theft, now put at over $6 billion across 11 blockchains, to North Korea's TraderTraitor, publishes 628 laundering wallets and asks firms to block them; Binance has started restoring withdrawals.
The FBI, Secret Service and IRS say stolen funds are being converted to Ether before further laundering through DeFi and instant-swap services; the amount is more than four times the Bybit theft of February 2025, The Hacker News reports.
severity critical · exploited in the wild · EU: MiCA, DORA, AMLR · actor TraderTraitor (90%), escalation

[P2] Apple notifications alert victims in 99 countries about mercenary spyware infections — CyberScoop
Why it matters: Apple has warned users in 99 countries, journalists and film-makers among them, of mercenary spyware infections, bringing the total to over 150 countries; it names no vendor or attacker.
Citizen Lab's John Scott-Railton calls the set of notifications significant, CyberScoop reports; the countries have not been listed.
severity high · exploited in the wild · EU: GDPR, EMFA

[P1] As Russia's Cyberwar Evolves, Belarus Steps Forward — darkreading
Why it matters: ESET reports the first destructive cyberattack it has documented in Europe by a Belarus-aligned actor: FrostyNeighbor (UNC1151), known for espionage and disinformation, hit a critical target in Ukraine.
The group has long targeted Ukraine and Poland through phishing; the report covers Q2 and Q3 2026 and suggests closer alignment with Russian aims, Dark Reading reports.
severity high · exploited in the wild · EU: NIS2 · actor FrostyNeighbor (UNC1151) (70%), escalation

[P3] New VOLTA Infostealer Uses Claude, Gemini, and Kimi AI to Rank Victims by Financial Value — The Hacker News
Why it matters: Cisco Talos finds VOLTA, an infostealer that feeds stolen data to Claude, Gemini or Kimi to rank victims by financial value and build a profile with home address, employer and birth year.
It also searches victims' Gmail for receipts to confirm addresses, using stolen sessions, The Hacker News reports; AI providers' abuse detection is one control point.
severity medium · exploited in the wild · EU: GDPR, AI Act

Health & Cyber

[P2] Crecen los ciberataques con agentes de IA en Argentina: vulneraron al menos 15 entidades de salud y robaron historias clínicas — AI Incident Database RSS Feed
Why it matters: An attacker using an autonomous AI pentesting framework compromised at least 15 Argentine health bodies between May and August and stole medical records from one, CrowdStrike research reported by Clarín shows.
The attackers left directories exposed that showed their prompts and the AI's actions; CrowdStrike did not name institutions or give patient numbers, Clarín reports.
severity high · exploited in the wild · EU: GDPR, NIS2, EHDS, AI Act

NHS quango warns of risk to GP patient data after website vendor collapse — www.theregister.com - Articles
Why it matters: NHS England warns GP practices about risks to patient information and service continuity after website provider Silicon Practice went into administration, leaving hundreds of practices affected and some sites offline.
CTO Kevin Humphries wrote to practices, which are independent contractors, to preserve data and keep patients informed, The Register reports. Vendor failure, not an attack, cut the digital channel to patients; the EHDS and NIS2 supply-chain rules treat continuity as a duty.

[P1] Hospital Cyberattacks Bring Disruptions, Legal Fallout — HealthcareInfoSecurity.com RSS Syndication
Why it matters: Cleveland's 21-hospital University Hospitals is still restoring systems after an attack on 22 October that took down its patient portal and forced rescheduled procedures, while UNM's attackers offered a free decryption key for hospital machines.
UNM services have resumed, and Byod claims 36,000 devices and said it would not have locked machines assisting patients in real time, HealthcareInfoSecurity reports. Separately, Lehigh Valley Health Network pays $800,000 over a 2023 attack that exposed photos of cancer patients.
severity high · exploited in the wild · EU: NIS2, GDPR, EHDS · actor Byod (UNM, self-claimed) (50%)

Who’s responsible when clinical AI goes wrong? AMA weighs in — Fierce Healthcare
Why it matters: The American Medical Association proposes a liability framework for clinical AI that spreads responsibility across developers, health systems and clinicians, warning that doctors now carry risks they cannot prevent.
It flags unclear product liability and contract disclaimers that separate responsibility from control, Fierce Healthcare reports. The EU's revised Product Liability Directive already treats software as a product, and the AI Liability Directive was withdrawn, leaving the same gap on fault.

Quantum & Cryptography

Alice & Bob Claims New Benchmark for Breaking RSA-2048 in Days — HealthcareInfoSecurity.com RSS Syndication
Why it matters: The authors of last week's estimate of breaking RSA-2048 with about 23,000 cat qubits are the French start-up Alice & Bob, whose figure is roughly 40 times below Google's 2025 estimate of one million qubits.
The company stresses it is a best case that depends on its own qubit architecture, HealthcareInfoSecurity reports, and such a machine does not exist. The claim comes from a firm selling that architecture, as Brussels prepares its Quantum Act. It adds urgency to the EU's post-quantum road map, but independent review of the estimate is still to come.

Cybersecurity & Threats

[P2] Official Claude Code Extension on Cursor Marketplace Trojanized with Persistent SSH Backdoor — The Hacker News
Why it matters: A trojanised update of Anthropic's Claude Code extension, under its legitimate identity, was served through Cursor's marketplace for about 12 hours to as many as 5,000 developers, trying to plant a persistent SSH backdoor.
Version 2.1.313 carried a 66-line loader under the ID anthropic.claude-code, only on Cursor's marketplace and not Open VSX or Microsoft's; Cursor shut the marketplace to remove it, and how attackers published it is unclear, The Hacker News and Ars Technica report. WIRED says flawed code stopped the backdoor from fully working. Anthropic says the compromise was not on its side.
severity high · exploited in the wild · EU: NIS2, CRA

[P1] Attackers Chain Three Ivanti Connect Secure Flaws to Gain Unauthenticated Root Access — The Hacker News
Why it matters: CISA lists two more Ivanti Connect Secure flaws as exploited, completing a chain that gives unauthenticated root access, used by a newly tracked actor to steal credentials and BitLocker recovery keys.
Volexity documented the chain (authentication bypass CVE-2026-104935, plus code execution and privilege escalation) used by UTA0576, which searched for administrator passwords and Active Directory metadata; Ivanti was still working on fixes when details came out, The Hacker News reports.
severity critical (CVSS 8.6) · exploited in the wild · CVE-2026-104935 · EU: NIS2 · actor UTA0576 (50%)

[P1] CISA says max severity GitLab flaw now exploited in the wild — BleepingComputer
Why it matters: CISA confirms exploitation of a maximum-severity GitLab flaw that lets unauthenticated attackers run code through a crafted regular expression; Alpine Linux rebuilt its GitLab server from scratch after attempts on 20 and 21 October.
CVE-2026-97229 is an integer overflow in the CI/CD regular expression compiler, fixed in 19.4.2, 19.3.5 and 19.2.8; Alpine found no tampering with packages but could not rule out compromise because logs were missing, BleepingComputer reports.
severity critical (CVSS 10.0) · exploited in the wild · CVE-2026-97229 · EU: NIS2, CRA

[P2] Chainalysis Confirms Intrusion Into Internal Tools and Exposure of Customer Identity Data — The Hacker News
Why it matters: Blockchain-analysis firm Chainalysis says an external actor used 'offensive agentic techniques' to reach its systems and internal tools this month, exposing customer names, emails and roles, but not investigation content.
The company gave no technical detail and says credentials and session tokens were not accessed, The Hacker News reports. Its customers include law enforcement and regulators, so even role data has value.
severity high · exploited in the wild · EU: GDPR, MiCA

[P2] Florida warns that 4.9 million people had Social Security numbers leaked in financial aid site breach — The Record from Recorded Future News
Why it matters: Florida says a breach of its student financial aid site exposed names, addresses and Social Security numbers of about 4.9 million applicants; the site is down.
The education department learned of the breach on 19 October and is working with state cyber and police units, The Record reports; aid payments are not affected.
severity high · exploited in the wild · EU: GDPR

[P2] New Android SMS of Death exploit can remotely crash unpatched phones — BleepingComputer
Why it matters: A researcher published a working exploit for an Android flaw that crashes phones or drops calls, including emergency calls, with a hidden text message; Android 12 and 13 remain unfixed.
CVE-2026-58882 is a zero-click denial of service from improper validation of a header length, processed before the message reaches the inbox; Google reportedly fixed it in October for Android 14 to 17, though it is not in the bulletin, BleepingComputer reports.
severity high · CVE-2026-58882 · EU: CRA, NIS2

The daily brief, every morning on Telegram →

Top Stories