Loading story…
Meta suspended its internal Model Capability Initiative after an internal error reportedly made sensitive employee monitoring data broadly accessible inside the company, intensifying criticism of a program that collected workplace activity from tools including Gmail, Google Chat, Meta AI, screenshots, and usage patterns. Meta said it found no evidence of malicious activity or an external breach, but the incident renewed concerns about consent, unencrypted storage, and how far companies are going to feed AI systems with employee data. Similar privacy concerns surfaced in the Netherlands, where a watchdog said the AIVD and MIVD intelligence services mishandled bulk datasets containing personal information through weak controls, excessive retention, and inadequate oversight, raising alarms that AI-assisted analysis could be applied to unlawfully sourced or improperly retained citizen data.
At the same time, new reporting and vendor changes underscored how quickly enterprise AI is expanding the attack surface and governance burden. Orca Security said 81.2% of organizations running AI packages had at least one known vulnerability and that 99.9% of fixable AI vulnerability alerts remained unpatched, with risks spanning agent frameworks, vector databases, exposed API keys, excessive permissions, weak authentication, and internet-facing services such as Amazon SageMaker, Azure OpenAI, and Google Vertex AI. Microsoft is adding an in-meeting Teams control to disable Copilot, Facilitator, and recap features during sensitive discussions, while broader industry debate has focused on employees oversharing data with AI assistants, the need to govern agent identities and registries, and warnings that AI prompts, memory, and orchestration layers can leak or concentrate proprietary organizational knowledge if enterprises do not retain tight control over access, telemetry, and data handling.
Mallory correlates global threat intelligence with your attack surface — know if you’re exposed before adversaries strike.
7 events from the most recent confirmed update back to the earliest known activity.
In 2022, the Dutch oversight body CTIVD issued a binding ruling ordering deletion of five major citizen datasets that had been retained too long by AIVD and MIVD. The ruling is cited as an earlier milestone in the broader controversy over intelligence data handling.
Microsoft said rollout of a new Microsoft Teams in-meeting toggle to enable or disable Copilot, Facilitator, and recap begins in early July 2026 for Targeted Release. The control is intended to give organizers and presenters real-time control over AI use during sensitive meetings.
Meta CTO Andrew Bosworth later said the exposure was caused by an internal error in which a researcher stored data in the wrong internal location, making it accessible more broadly than intended. This clarified the company’s attribution of the incident as an internal mistake rather than an outside compromise.
Meta suspended its internal Model Capability Initiative after discovering that sensitive employee information had been made broadly accessible inside the company. Meta opened an internal investigation and filed an internal security incident, saying it found no evidence of malicious activity or an external breach.
More than 1,600 Meta employees signed a petition calling for the Model Capability Initiative to end after criticizing it over privacy, consent, and reports that some collected data was stored unencrypted. The backlash preceded Meta’s later suspension of the program.
Meta launched the internal Model Capability Initiative in April to collect workplace activity data for improving its AI systems. The program monitored tools including Gmail, Google Chat, and Meta’s AI assistant, and captured screenshots and usage patterns.
A CTIVD supervisory report dated July 1 found that the Netherlands’ intelligence services mishandled bulk datasets through weak technical controls, excessive retention, incomplete access restrictions, and insufficient oversight. The watchdog issued thirteen recommendations for stronger logging, clearer legal definitions, and better dataset lifecycle management.
Vulnerabilities, threat actors, malware, products, organizations, and breaches Mallory has linked to this story.
9 references tracked. Mallory keeps watching after this page renders.
teiss.co.uk
Open sourcethenewstack.io
Open sourcethecybersecguru.com
Open sourcetechtarget.com
Open sourcevirtual-routes.org
Open sourcehelpnetsecurity.com
Open sourceghacks.net
Open sourcecysecurity.news
Open sourcesemperis.com
Open sourceMap indicators from this story to your assets and identify affected systems in minutes.
Every observed campaign, victim, and pivot linked to actors named in this story.
Malware, exploits, and IOCs connected to the activity described here.
YARA, Sigma, and Snort rules deployed to your SIEM as soon as they’re published.
Get matching new stories delivered to your team as they break — not the next morning.
Ask questions about this story and take action on the answers.
Other stories Mallory is tracking that overlap on entities, sources, or topic labels.
Reporting focused on **AI’s accelerating integration into products and workplaces** highlighted two primary risk areas: privacy exposure from AI features that process user data off-device, and organizational strain as companies and employees adapt to AI-driven change. Commentary and analysis emphasized that current large language models still exhibit core reliability issues—*hallucinations*, uncertainty, and overconfidence—while investment and infrastructure buildout for AI data centers continues to surge. A WhatsApp Android beta feature that uses **Meta AI** to help users organize chat history drew renewed scrutiny because prompts and responses are processed on Meta’s servers and fall outside WhatsApp’s standard end-to-end encrypted user-to-user model; the feature also introduces “threaded” AI conversations while potentially retaining shared “memory” across threads unless users disable it, raising **data handling and consent** questions (notably in the EU). Separately, workplace-oriented pieces described growing anxiety about job displacement and skills erosion as AI adoption expands, with guidance urging workers to adapt and opinionated commentary warning that reliance on AI-generated code can reduce engineering comprehension and increase operational risk if experienced reviewers are cut or lack system-level understanding.
Jun 29, 2026
Security researchers and enterprise administrators are warning that connecting AI assistants such as **Claude**, **ChatGPT**, and **Copilot** to business platforms like Microsoft 365, Gmail, Slack, Dropbox, Zoom, Outlook, SharePoint, and OneDrive can sharply expand the attack surface. Reporting on the connector ecosystem said hundreds of integrations changed within weeks and many can pass data to additional AI services or subprocessors, complicating governance and creating hidden exposure outside primary enterprise controls. Anthropic documentation also notes that connected services process data on their own infrastructure and under their own terms, adding compliance concerns for organizations handling sensitive or regulated information. The operational risk is not limited to data access: security coverage described **"authority laundering"**, where untrusted input is transformed by one AI system into trusted instructions for another system authorized to act, including a cited case involving fund movement. Sysadmin discussions echoed those concerns, with organizations debating whether to allow broad OAuth access to Outlook, Teams, calendars, files, and even local device control, especially in HIPAA-regulated environments. Microsoft is meanwhile replacing Outlook's confusing **Meeting Insights** feature with a deeper **Microsoft 365 Copilot** meeting-prep experience, while vendors such as Ledger are promoting safer agent designs that keep AI actions read-only until a human approves sensitive operations on hardware.
Jul 19, 2026
Enterprises’ rapid deployment of **AI and agentic AI** is increasingly creating measurable security and business risk, including direct exposure of sensitive personal data and downstream impacts on risk transfer. A widely cited example involved McDonald’s *McHire* applicant-screening platform (built by *Paradox.ai*), where researchers reported a trivial backend credential weakness (`123456` as both username and password) and no MFA, potentially exposing data tied to roughly **64 million** applicants; the incident is being used by insurers and risk teams as evidence that AI adoption is moving faster than security and governance, contributing to tighter cyber-insurance language, higher premiums, and **AI-related exclusions**. Separate reporting also highlighted that “plug-and-play” AI is unrealistic at enterprise scale, with organizations increasingly needing custom integration and operational ownership rather than relying on off-the-shelf tools. Threat reporting during the same period reinforced that AI is expanding both attacker capability and the attack surface: researchers described **Pakistan-linked APT36** using AI coding tools to generate high volumes of low-quality malware variants (including in less common languages) and to leverage legitimate cloud services for command-and-control, complicating detection. Additional research flagged **AI-themed browser extensions** (Chrome/Edge) that impersonate legitimate tools and can harvest LLM chat histories and browsing activity, underscoring the risk of “shadow AI” and unvetted add-ons. In parallel, routine threat-intelligence summaries continued to track major incidents (e.g., ransomware and data breaches) alongside AI-enabled tactics, indicating that AI risk is becoming intertwined with broader enterprise security exposure rather than remaining a standalone technology concern.
Jun 29, 2026