Source code stories
Detection of malicious code can collapse when AI reviewers are fed large files packed with harmless text, Cloudflare's research shows.
Recent AI-driven leaks are forcing firms to rethink IP protection as sensitive code and creative assets move across cloud tools and public repositories.
Engineering teams can now keep decisions, fixes and costs in one place as CodeRabbit brings its AI agent into Slack.
A critical flaw in a widely used Microsoft code-sample repository could have let attackers steal secrets and run code through GitHub issues.
Enterprises under release pressure can now test more quickly, as Leapwork combines functional automation, performance testing and AI orchestration in one platform.
Security researchers say long automated jobs can make Claude Code’s deny rules fall back to user prompts, weakening protections in CI/CD pipelines.
Sensitive prompts and documents will stay out of model training as ExpressVPN enters AI software with an enclave-based service for Pro subscribers.
Sysdig unveils runtime security for AI coding agents, promising real-time monitoring of autonomous dev tools to curb emerging risks.
SpecterOps broadens BloodHound Enterprise to map identity attack paths across Okta, GitHub and Jamf-managed Macs in hybrid environments.
Secure Code Warrior launches SCW Trust Agent: AI, giving security teams commit-level visibility and control over AI-influenced code.
AI-fuelled coding drives record 29 million hardcoded secrets on GitHub in 2025, with leaks from AI tools and services surging sharply.
Checkmarx overhauls its One platform with AI-native security agents to guard fast-moving, agentic development and AI software supply chains.
MIND launches Autonomous DLP Analyst to automate classification and investigation, cutting noisy alerts and easing data security workloads.
A flaw in a Microsoft GitHub workflow could let attackers run unauthorised code and steal repository secrets, Tenable said.
It could cut months from modernisation projects by turning undocumented legacy code into design documents, with Fujitsu already trialling the tool at banks.
Australian organisations face fresh risk of cloud and identity compromise as the cyber watchdog reissues its alert on repository attacks.
AI disruptions and cyberattacks are forcing organisations to back up models, prompts and knowledge bases, not just files.
Users can now query AI without prompts or files being exposed, as ExpressVPN moves beyond virtual private networks into confidential computing.
The new suite could ease Europe’s reliance on Microsoft Office, as a coalition opens its code ahead of a summer stable release.
Backslash adds cross-tool governance to discover, vet and monitor 'Skills' powering AI coding assistants like Cursor, Claude Code and Copilot.
