TokenPost topic
cybersecurity
643 TokenPost stories are tagged cybersecurity. The newest appear first.
Latest cybersecurity stories
- AI May Speed Crypto Security Reviews as OpenAI Math Results Show No New Weakness
A model’s monthlong test would strengthen confidence only if it could not break the assumption under review.
· Technology
- Bank of America Keeps 1,000 Contractors on Mythos-Linked Repairs
The bank spent more than $200 million in the third quarter on software remediation and anti-money-laundering matters combined, without breaking out Mythos costs.
· Technology
- OpenAI Hires Graham Seamans to Lead Cybersecurity Sales
Seamans previously led sales at xAI and worked at SpaceX as OpenAI expands its enterprise and cybersecurity offerings.
· People
- Agentic Tools Cut Attack-to-Impact Time to Hours in One Campaign Against Indian Entities
The timeline shrank from days to a few hours. In separate AI evaluations, agents reached live systems and read records containing personal data and confidential material.
· Technology
- G7 Finance Officials Plan Tabletop Exercises on AI Risks
Japanese Finance Minister Satsuki Katayama announced the plan after G7 finance ministers and central bank governors met in Bangkok on Oct. 15.
· Regulation
- AI Agents Under Test Accessed Live Systems at Indian Companies
AI model providers voluntarily disclosed each case after agents exceeded their task authorizations; affected companies did not detect the access.
· Technology
- Jay Clayton Urges U.S. Firms and Government to Be Wary of AI-Era Spying
Clayton tied the warning to AI-amplified cyber risks to interconnected infrastructure and said the task force established three subcommittees at its first meeting Thursday.
· Regulation
- OpenAI Model Found and Used a Sepolia Wallet to Deploy a Smart Contract
It checked 362 addresses derived from public developer recovery phrases and selected one with enough test Ether; the deployment succeeded and fees were charged to that account.
· Technology
- AI Agents Bypassed Isolation Controls in July Hugging Face Breach
The reconstructed timeline covered about 17,600 actions from July 8-13 ET; five datasets were the only customer content accessed.
· Technology
- Survey Finds AI Involved in 84% of Incidents at Ransomware-Hit Firms
The survey covered organizations hit by ransomware in the prior year. Identity systems were targeted in 87% of incidents, while nearly 70% of victims faced repeat attacks.
· Technology
- Zeabur Discloses Second Breach in Seven Weeks, Exposing API Tokens
The company said the intrusion also reached some user-rented dedicated servers on Google Cloud and urged customers to replace exposed credentials.
· Technology
- Trump’s AI Task Force Agrees to Address Three AI Risk Areas
At its Oct. 15 meeting, the task force planned a report assessing those risks and other potential concerns.
· Regulation
- Proofpoint Agrees to Acquire Varonis in $6.3 Billion Cash Deal
Varonis holders would receive $49.75 per share. The deal needs shareholder and regulatory approval and is targeted to close in the first half of 2027.
· Business
- KPMG U.S. CEO Says AI Is No Y2K as 53% Plan Cuts in Some Areas
A survey of 400 U.S. CEOs found that 89% expect total headcount to grow over three years, even as many plan reductions in some areas over two to five years.
· Business
- Polygon Labs Releases Bor v2.10.3 to Fix Two Medium-Severity Vulnerabilities
The flaws affected Bor v2.10.1 and v2.10.2, where BlockSTM V2 was enabled by default. The update does not require a hardfork.
· Technology
- Cohere Signs Letter of Intent to Acquire AI Cybersecurity Startup AISLE
AISLE develops AI tools to identify and help repair software vulnerabilities.
· Business
- South Korea’s KOSPI Falls 7.57% as Financial Regulator Reviews Risks
A separate cybersecurity review found breaches at 16 financial firms and about 100,000 exposed pieces of customer data.
· Investing
- Proofpoint Plans Four Microsoft 365 Security Tools for MSPs in Q4
The planned additions span tenant access monitoring, email assessment explanations, AI-assisted Teams training and an administrator agent.
· Technology
- South Korea Confirms Breaches at 16 Financial Institutions
About 100,000 personal-information records were leaked. Affected institutions include Streami, operator of crypto exchange GOPAX, and virtual asset service provider Prabang.
· Technology
- South Korea Plans More Frequent Cybersecurity Checks After 16 Breaches
About 100,000 pieces of customer data were exposed, while a sectorwide checklist flagged security deficiencies at 130 of 555 financial companies.
· Regulation
- Coinbase’s Cold Storage Shifted From USB Drives to Key Ceremonies
A 2012 account put 87% of customer funds offline, while later descriptions detailed Faraday tents and controls for generating and recovering keys.
· Technology
- AI Researchers Call for Internet Isolation of Cyber-Capable Models
A letter dated October 2026 proposes an initial airgap default and an international body to set containment standards and track failures.
· Technology
- Microsoft and Japan Disrupt Network Behind 1 Billion Scam Messages
The campaigns targeted more than 75,000 organizations and people in more than 150 countries. Microsoft disabled more than 20,000 accounts tied to YAM Store.
· Technology
- OpenAI CFO Urges Companies to Use More Western Models for Cybersecurity
Sarah Friar also said Chinese AI models had become strong in cybersecurity. She did not name specific models.
· Technology
- Bank of America Hires 1,000 Contractors to Patch AI-Exposed Software Flaws
About 3,000 employees joined the effort. Remediation and anti-money-laundering compliance added hundreds of millions of dollars in third-quarter costs.
· Technology
- SentinelOne Executive Says Daybreak Cut Malware Analysis to Eight Hours
Gregor Stewart described one reverse-engineering task that took about eight hours, compared with four to six weeks for a specialist. He said human accountability remains part of the process.
· Technology
- Nvidia CEO Jensen Huang Says AI Industry Should Not Tolerate Cybersecurity Incidents
Huang said stronger containment and monitoring could have prevented AI agent incidents. Nvidia plans to announce $2 billion in new European investments within the next week.
· Technology
- Anthropic Tests Two Steps Toward an AI Worm, Not Real-World Infection
Tests ran a model across 15 Android TV boxes and simulated replication among cloud-hosted Android devices. They did not demonstrate a worm infecting real-world devices.
· Technology
- Bill Gates Warns AI Could Spur More Cyberattacks Over Next Year
He cited extortion and efforts to move stock and betting markets as risks, and urged governments to develop more sophisticated oversight and safety groups.
· Technology
- Citi Raises ServiceNow Price Target to $170 as AI Interest Improves
The revised target is up from $154 and implies 22% upside. ServiceNow shares gained about 1% in early Wednesday trading.
· Business
- Submarine Cables Carry More Than 99% of Global Data Flows
Author Elisabeth Braw’s book examines risks to undersea infrastructure as routine damage and geopolitical tensions complicate investigations into cable disruptions.
· Technology
- CodeWall Emerges From Stealth With $4 Million in Pre-Seed Funding
The London cybersecurity startup is developing AI agents that test enterprise systems and retest fixes within customer-approved limits.
· Business
- XBOW Earns $250,000 Google Bounty for Chrome Exploit
The six-day effort chained three Chrome vulnerabilities on Windows to run code outside the browser’s sandbox. XBOW paired AI agents with a human security researcher.
· Technology
- Ledger Orders Southeast Asian Dealers to Halt Sales and Recall Devices
The company is investigating malicious components found in devices sold by CryptoBilis, with recalled units to be examined by Ledger Donjon.
· Technology
- Patch the Planet Identifies Hundreds of Open-Source Bugs in First Week
OpenAI and Trail of Bits launched the cybersecurity initiative under OpenAI’s Daybreak program on June 22, 2026, with work spanning widely used software projects.
· Technology
- Uber CISO Urges Real-Time Controls for Frontier AI Systems
Philip Martin said advanced models should face oversight similar to employee insider-threat programs as they process more context and act beyond expected constraints.
· Technology
- GitHub Cuts Secret-Scanning Time 86% With ModernBERT Classifier
The context-aware system increased repository-owner-weighted acceptance by 11.5% and reduced serving costs by more than 98% compared with a remote large-language-model pipeline.
· Technology
- Internet-Exposed AI Infrastructure Climbs 63% in Nine Months
More than 294,000 public IP addresses exposed 43 AI and large-language-model tools, including platforms affected by serious vulnerabilities.
· Technology
- SentinelOne Plans $1.25 Billion Convertible Note Placement
The senior convertible notes would mature in 2031. SentinelOne has also raised fiscal 2027 revenue guidance to $1.202 billion-$1.207 billion.
· Business
- GLM-5.3 Nearly Matches Anthropic Model in Cyber Exploit Tests
The open-weight model produced working exploits in 50 of 410 attempts, while U.S. testing ranked it the most capable open-weight model evaluated to date.
· Technology