Skip to main content
AI-Intel.news logoAI-Intel.news

Section

Security

Vulnerabilities, privacy, data governance, and AI risk.

492 published stories

Follow Security to track important changes in this topic. Not every new development will appear — only material change.

What AI security coverage tracks

AI security spans two directions at once: attacks that target AI systems, and attacks that use AI to move faster. This section follows both, from prompt injection and agent misuse through model supply-chain weaknesses, data exposure, and AI-assisted intrusion and fraud campaigns.

Each item here is compiled from primary sources — vendor advisories, research publications, regulator and law-enforcement notices, and company disclosures — and grouped into ongoing Developments so a single incident and its follow-ups stay together instead of arriving as unconnected headlines.

What you'll find in this section

  • Vulnerabilities and incidents affecting models, agents, and AI infrastructure
  • Attacker use of AI for phishing, fraud, malware, and reconnaissance
  • Vendor advisories, patches, and mitigation guidance as it is published
  • Security requirements arriving through regulation and procurement rules

Development Intelligence

Key developments

5 developments in Security where several reports describe the same story.

DevelopmentDevelopingSources: 4 independent sources + 1 further report

Sam Altman and Elon Musk support Dario Amodei's call for AI development slowdown

Financial Times reports that rival AI leaders Sam Altman and Elon Musk have supported Anthropic CEO Dario Amodei's call for a development slowdown, uniting around shared warnings that humans could lose control of advanced artificial intelligence. Claims are as reported; this summary makes no determination about accuracy or significance.

  • Moderate confidence
  • Strongly corroborated

Coverage

View development →
DevelopmentDevelopingSources: 3 independent sources

Apple tightens macOS Full Disk Access controls due to AI agent risks

Apple announced plans to tighten macOS Full Disk Access controls, stating that increasingly capable AI agents elevate the risks of granting broad permissions across sensitive user data, such as local files, email, messages, and browsing history. Claims are as reported; this summary makes no determination about accuracy or significance.

  • Very high confidence
  • Corroborated

Coverage

View development →
DevelopmentSources: Primary source + 7 independent reports

Meta launches personal assistant AI agent Muse

The Verge reports that Meta is launching Muse, a personal assistant AI agent aimed at broad consumer adoption. Claims are as reported; this summary makes no determination about accuracy or significance.

  • Very high confidence
  • Widely corroborated

Coverage

View development →
DevelopmentSources: Primary source + 7 independent reports

OpenAI launches Astra model

TechCrunch reports that OpenAI has launched Astra, a new model designed for computer and browser use. Claims are as reported; this summary makes no determination about accuracy or significance.

  • Very high confidence
  • Widely corroborated

Coverage

View development →
DevelopmentSources: Primary source + 3 independent reports

OpenAI reports Astra model crosses Critical cybersecurity capability threshold

OpenAI says its upcoming Astra AI model is its first to reach a "Critical" cybersecurity capability threshold, according to CNBC. The company stated that Astra will be released soon, though access to its cybersecurity capabilities will be restricted. Claims are as reported; this summary makes no determination about accuracy or significance.

  • Very high confidence
  • Widely corroborated

Coverage

View development →

Latest coverage

The InformationEnterprise

Salesforce Unveils New AI Model, Agent Security Tools at Dreamforce

Salesforce unveiled a new AI model named Koa and introduced security tools designed to help enterprise customers run AI agents safely on corporate data, according to reporting from Dreamforce by The Information.

41/100Intel Score, moderate impact
The InformationBusiness

Why Anthropic’s Data Policy Drama Is Good for OpenAI

Major enterprise customers including Palantir and Booz Allen are restricting their use of Anthropic's newest Fable AI models and turning to OpenAI's Astra model. The shift follows Anthropic's June policy change requiring 30-day data retention for security reasons on Fable models, while OpenAI provides zero data retention options.

48/100Intel Score, moderate impact
The DecoderBusiness

AI labs have a data trust problem that their policies haven't solved

Anthropic's policy to store usage logs for 30 days from its flagship model Fable prompted enterprise customers including Palantir, Nvidia, and Booz Allen Hamilton to withdraw the model from sensitive workloads, highlighting persistent corporate data trust concerns despite lab assurances against training on customer data.

67/100Intel Score, high impact
CNBC TechBusiness

Trump’s opposition to AI rules undercuts industry's calls for a slowdown

CNBC reports that political opposition from Donald Trump against AI regulation conflicts with appeals from frontier labs, including Anthropic and OpenAI, which have requested mandatory safety evaluations amid rising concerns over controlling advanced artificial intelligence systems.

48/100Intel Score, moderate impact
CIO DiveEnterprise

Security teams increasingly outflanked by AI agents

CIO Dive reports that enterprise security teams face tracking challenges as non-human identities expand rapidly alongside AI agent deployment, citing findings from IDC and GuidePoint Security showing traditional tracking systems are being outpaced.

64/100Intel Score, high impact
The DecoderBusiness

OpenAI has hundreds of contract workers reading your ChatGPT conversations

OpenAI employs hundreds of contract workers to review and rate real ChatGPT conversations on a one-to-seven scale to reduce flattery and overly human-like behavior, according to reporting from 404 Media. While prompt text is anonymized before contractor review, the conversations can still contain sensitive user information. Human review is enabled by default under the "Improve the model for everyone" setting, requiring users to actively opt out to prevent contractor access.

64/100Intel Score, high impact
The InformationBusiness

Anthropic Data Fears Prompt Nvidia, Palantir and Booz Allen to Restrict Model Use

Major enterprise and defense contractors, including Palantir Technologies, Nvidia, and Booz Allen Hamilton, are restricting or eliminating their use of advanced models from Anthropic and OpenAI, according to The Information. The actions stem from heightened concerns that proprietary customer intellectual property could be absorbed or learned by foundation models, prompting these organizations to demand stronger data guarantees.

55/100Intel Score, high impact
Financial Times (AI)Regulation

China’s spy agency warns of AI risk to national security

Financial Times reports that China's state intelligence apparatus has issued a warning highlighting national security risks posed by artificial intelligence. The alert indicates heightened concern within Beijing that rapid advances in AI capabilities could be weaponized or deployed against the state.

46/100Intel Score, moderate impact