IT
Agent frameworks, SDKs and CLIs, open source releases, local inference and coding tools.

10 security checks for AI-built apps before sign-up
Testing whether an app’s features work and whether it can safely handle real user data are separate jobs.
Last 7 daysNov 17 – Nov 21
Grok Bot requests access to company data without new prompts
NewCore reported that all the requests expired without approval.
Web course covers AI agent creation and control
The course presents experiments with real agent interactions as a way to learn permissions, security, monitoring and collaboration.
Coinbase presents AI agent trading as a revenue source
The trading volume forecast sits alongside restoring trust as a stated condition for Coinbase's expansion.Fasoo AI puts agent sourcing at centre of AI strategy
Fasoo AI frames the choice between building and buying agents as part of AI transformation strategy for businesses and institutions.
Forter plans AI payment security showcase at G-Star 2026
Forter plans to discuss security applications and cooperation with game companies, drawing on risk analysis from account creation through activity after purchase.Agent memory test highlights risks from stored secrets
A fake-key test raises two linked issues: removing secrets before storage and their later reuse through agent memory.
LG Uplus adds AI data controls to network security
U+SASE promotes controls on AI inputs, but its cited school deployments cover web access blocking.
Voice and email spam reports rise in first half
The user survey of spam received and the tally of reports and detections measure different things and need to be read separately.
Trusted, connected AI agents need distinct security
A DEV post challenged the security approach of treating capable, connected AI agents as ordinary software.
Hanjin IT arm plans AI push into external markets
Hanjin's IT services arm planned to commercialise its aviation and logistics expertise through AI, supported by stronger cloud and security capabilities.
AI design approach puts control into system architecture
A DEV post presented a design approach that treats purpose and control rules as executable parts of an AI system's architecture.
1 week agoNov 10 – Nov 16
Postman drill examines where agent-handled tokens end up
A staged drill using a fake token shifts attention from whether an agent completes a replacement to where secrets remain.
BlackRock accounts for about 90% of bitcoin ETF outflows
One fund's share of total daily net outflows and its consecutive outflow streak measure different things.
Blockstream puts custody signing inside security hardware
Blockstream's design runs the entire signing process inside security hardware, according to the company account reported by DigitalToday.
Elecom’s new routers support Wi-Fi 7 on the 5GHz band
Wireless speed limits apply to individual bands, while the security subscription included with one model is free for one year.
Meta Muse request share does not measure web market share
One security firm's observations show AI agent activity, but do not establish user numbers or purchase outcomes.
Arlo Essential Security Camera HD gets early discount to $10
The $10 price reported by XDA comes with an early-discount condition.
Lotte Innovate links AI monitoring expansion to web security
The Elec reported a plan to connect web protection to Lotte Innovate’s own security monitoring platform and extend integrated services to external businesses.
Stopping the model can leave an AI app running
Tasks that can persist after model shutdown show why containment testing needs to cover the whole AI app.
Replit users bear responsibility for Agent-generated code
A security checklist described Replit’s division of responsibility between users who own generated code and the company that protects the platform.
Anthropic models differ in research request restrictions
Reportedly identical core capabilities do not mean the two models allow the same biology and cybersecurity requests.
AI SPERA launches malicious IP matching app for Splunk
The app automates comparisons between threat data and corporate firewall logs, a function that should be assessed separately from detection effectiveness.
Lotte Innovate plans AI security service with Thales tech
Lotte Innovate plans to connect web protection to its monitoring platform and extend integrated security services beyond Lotte affiliates.
IBM and Microsoft unveil enterprise AI agent platform
Imagine Enterprise centres on adapting pre-built assets to company workflows, while its customer example focuses on technology exploration.
Synap Document Viewer adds AI translation that keeps layout
The update adds translation that preserves document formatting, while the share of features requested by customers describes development priorities rather than translation quality.
Support AI launch puts action permissions under scrutiny
A favourable penetration test report needs to be read alongside the agent’s permission to change customer records and issue refunds.
MobiScore wins hackathon award
Mobiligent plans paid pilots with transport operators before commercialisation, while vehicle data remains limited to internal testing.
Amazon launches Ring Drone Cam at $250
Ring Robotics launches with two devices, but the drone camera's reported price does not apply to the whole range.
Anthropic faces formal data exposure investigation in Korea
Anthropic's new model launch emphasises capability and safety, while Korea's privacy regulator investigates the scope of business account data exposure and how internal systems were accessed.
2 weeks agoNov 03 – Nov 07
Hugging Face's transformers v5.22.2 hardens Jinja sandboxing
The patch hardens Jinja sandboxing, but the release notes do not name the specific flaw it was meant to block.
Amazon Quick lets a separate model approve its agent actions
Smart Approvals is on by default, so AWS advises administrators to check first which actions stay under human approval.
CWE-bench v2 opens 100 security tasks to AI model developers
Even the leading model failed 19% of the tasks on every attempt, and the announcement gives no licence or commercial terms.
3 weeks agoNov 01
OpenAI fixes ChatGPT account takeover flaw set off by a link
Parameter reproduced a path that needed only one opened link, but its post does not say whether the flaw was exploited in practice
Oct 2026Oct 12 – Oct 22
Claude Code v2.1.309 fixes sandbox and permission check gaps
A flaw that ran shell startup files outside the sandbox is fixed, and the release also changes behaviour that affects automation scripts.
NVIDIA says AI agent controls must operate outside the agent
NVIDIA said monitoring and enforcement need to operate beyond an AI agent's reach because an agent that completes its task can still cross a security boundary.
Cursor opens Origin app marketplace to developer tool makers
Origin now has more than 20 integrated apps and will review outside developer tools for listing, but the changelog gives no review time or listing fee.
Devin Desktop 3.12.28 brings per-session MCP server controls
The release also fixes local-session and Remote SSH errors, but the changelog lists no vulnerability numbers for its security fixes.
38 in totalWritten by Jiyul SeoRSS feed
