Skip to content
cisonews.ai

Browse by year · 30 regulatory milestones · 15 incidents

2025 on cisonews.ai

Everything we have on record from 2025. Each page sits under the month its milestone or incident happened, newest first. cisonews.ai launched on September 27, 2026; entries about earlier events were compiled after that from public records and are labelled that way.

December 20256 entries

  1. Regulatory milestoneChinaChina's CAC drafts rules requiring human-like AI to disclose it is not humanCAC circulated draft Interim Measures that would force AI companions and chatbots to disclose they are AI, not a person.
  2. Regulatory milestoneUS StatesNew York signs the RAISE Act, a frontier AI safety and transparency lawGovernor Hochul signed the RAISE Act on December 19, 2025, requiring frontier AI developers to publish safety frameworks and report incidents within 72 hours.
  3. Regulatory milestoneUS FederalNIST releases preliminary draft of the Cyber AI ProfileNIST IR 8596 proposes a Cybersecurity Framework profile for managing AI system risk and AI-enabled attacks and defense.
  4. Regulatory milestoneUS FederalExecutive Order 14365 targets state AI laws with a national policy frameworkEO 14365, signed 11 December 2025, set up a Justice Department AI Litigation Task Force and ordered Commerce to list onerous state AI laws within 90 days.
  5. Regulatory milestoneUS FederalOMB tells agencies how to buy LLMs under the 'Unbiased AI' executive orderM-26-04 requires agencies to get vendor transparency documents and add compliance clauses to LLM contracts, implementing EO 14319's two Unbiased AI Principles.
  6. Regulatory milestoneInternationalCISA, NSA and Partners Publish AI-in-OT Security PrinciplesCISA, NSA, ASD's ACSC and allied cyber agencies issue four principles for securely integrating AI into operational technology environments.

October 20251 entry

  1. Regulatory milestoneUS StatesCalifornia signs SB 243, the Companion Chatbot Law, regulating AI chatbotsCalifornia's SB 243 requires AI companion chatbot operators to disclose non-human status, protect minors, and report suicide-related safety data.

September 20254 entries

  1. Regulatory milestoneUS StatesCalifornia SB 53, the Transparency in Frontier Artificial Intelligence Act, signedGovernor Newsom signed SB 53 on 29 September 2025, requiring large frontier AI developers to publish safety frameworks and creating incident reporting to Cal OES.
  2. IncidentSupply chainMalicious postmark-mcp npm package forwarded user emails to an attackerA backdoored npm package impersonating Postmark's MCP server secretly BCC'd outgoing emails to an attacker before being pulled from npm.
  3. Regulatory milestoneUS StatesCalifornia finalizes CCPA rules on automated decisionmaking technologyCPPA's ADMT, risk-assessment and cybersecurity-audit regulations under the CCPA cleared final state review on September 22, 2025.
  4. IncidentModel abuseAnthropic reports AI-orchestrated cyber espionage campaign run through Claude CodeAnthropic said in November 2025 that a group it linked to China used Claude Code to attack about thirty targets in mid-September 2025, succeeding in a few cases.

August 20255 entries

  1. IncidentModel abuseAnthropic discloses vibe hacking AI data extortion campaign (GTG-2002)Anthropic's August 2025 Threat Intelligence Report details how Claude Code was misused to run a large-scale data extortion operation tracked as GTG-2002.
  2. IncidentSupply chainNx s1ngularity attack: malicious npm packages turned local AI tools toward secret huntingOn 26 August 2025 attackers published malicious Nx packages to npm that scanned for credentials, tried local AI tools and posted results to public GitHub repos.
  3. IncidentPrompt injectionGitHub Copilot and Visual Studio prompt injection code execution flaw (CVE-2025-53773)CVE-2025-53773, published on 12 August 2025, is a command injection flaw in GitHub Copilot and Visual Studio that let an attacker run code on a developer machine.
  4. IncidentData exposureShared ChatGPT conversations found indexed on Google search in 2025OpenAI pulled a discoverability option after shared ChatGPT links began turning up in Google search results, exposing sensitive conversation details.
  5. Regulatory milestoneEUEU AI Act obligations for general-purpose AI models start to applyOn 2 August 2025 the AI Act's governance rules and duties for general-purpose AI model providers became applicable, with AI Office enforcement phased in later.

July 20255 entries

  1. Regulatory milestoneEUEuropean Commission publishes guidelines for providers of general-purpose AI modelsThe Commission's July 2025 guidelines set FLOPs thresholds and compliance criteria for GPAI providers under the EU AI Act.
  2. IncidentSupply chainMalicious wiping prompt shipped in Amazon Q Developer extension for VS CodeIn July 2025 an attacker used an over-scoped GitHub token to slip wiping instructions into Amazon Q Developer for VS Code version 1.84.0, which AWS then pulled.
  3. Regulatory milestoneEUEU publishes the General-Purpose AI Code of Practice under the AI ActThe final GPAI Code of Practice was published on 10 July 2025, a voluntary route for model providers to show AI Act compliance on transparency, copyright and security.
  4. IncidentPrompt injectionCursor AI code editor patches CurXecute prompt-injection RCE flawAimLabs found a single externally hosted prompt could hijack Cursor's AI coding agent into executing attacker commands with developer-level privileges.
  5. IncidentData exposureMcHire AI hiring chatbot exposes McDonald's applicant dataWeak admin credentials and an IDOR flaw in the Paradox.ai-built McHire chatbot exposed McDonald's job applicant records.

June 20253 entries

  1. IncidentPrompt injectionEchoLeak: zero-click prompt injection flaw in Microsoft 365 Copilot (CVE-2025-32711)EchoLeak, disclosed on 11 June 2025, let a crafted email trick Microsoft 365 Copilot into leaking internal data with no user click. Microsoft fixed it server-side.
  2. IncidentData exposureAsana's MCP server bug exposed data across customer organizationsA bug in Asana's MCP server, discovered June 4, 2025, could have exposed one customer's data to other Asana MCP users, CSO Online reported.
  3. Regulatory milestoneUS FederalUS AI Safety Institute renamed Center for AI Standards and InnovationCommerce Secretary Howard Lutnick announced the AI Safety Institute would become CAISI, shifting its focus from safety to standards and evaluation.

May 20253 entries

  1. Regulatory milestoneJapanJapan's National Diet enacts the Act on Promotion of AI-Related TechnologyJapan passed its first dedicated AI law on 28 May 2025, a non-binding, pro-innovation framework with no penalties for noncompliance.
  2. Regulatory milestoneUS FederalTAKE IT DOWN Act signed into US law, targeting deepfake intimate imagesFederal law criminalizes nonconsensual intimate images, including AI deepfakes, and sets a 48-hour platform takedown duty.
  3. IncidentDeepfake fraudFBI warns of AI-generated voice messages impersonating senior US officialsOn 15 May 2025 the FBI warned that actors were sending texts and AI-generated voice messages posing as senior US officials to gain access to accounts.

April 20252 entries

  1. Regulatory milestoneUS FederalOMB replaces federal AI memos with M-25-21 and M-25-22On 7 April 2025 the White House announced OMB memos M-25-21 on federal AI use and M-25-22 on AI acquisition, replacing the earlier approach to agency AI.
  2. Regulatory milestoneInternationalETSI publishes TS 104 223, baseline cyber security requirements for AIETSI's Securing Artificial Intelligence committee publishes TS 104 223 V1.1.1, setting lifecycle-wide baseline security requirements for AI models and systems.

March 20252 entries

  1. Regulatory milestoneUS FederalNIST publishes 2025 adversarial machine learning taxonomy, AI 100-2 E2025NIST published AI 100-2 E2025 on 24 March 2025, a taxonomy and glossary of attacks on predictive and generative AI and of mitigations against them.
  2. Regulatory milestoneEUEU adopts the Scientific Panel Implementing Act under the AI ActEuropean Commission adopts Implementing Regulation (EU) 2025/454, setting up a panel of independent experts to support AI Act enforcement.

February 20254 entries

  1. IncidentModel abuseMicrosoft names Storm-2139 developers who abused Azure OpenAI ServiceMicrosoft's Digital Crimes Unit amended a lawsuit to name developers accused of using stolen credentials to bypass Azure OpenAI safeguards.
  2. Regulatory milestoneEUEuropean Commission publishes guidelines on the AI Act's AI system definitionOn 6 February 2025 the Commission issued non-binding guidance on the seven-part test for what counts as an AI system under the AI Act.
  3. Regulatory milestoneEUEuropean Commission publishes guidelines on the AI Act's banned AI practicesThe Commission's February 2025 guidelines explain how the EU AI Act's bans on manipulation, social scoring and biometric surveillance apply.
  4. Regulatory milestoneEUEU AI Act bans on prohibited practices and AI literacy duty start to applyFrom 2 February 2025 the AI Act's prohibited practices and its AI literacy obligation applied across the EU, the first of the law's staged deadlines.

January 202510 entries

  1. Regulatory milestoneUKUK publishes the AI Cyber Security Code of PracticeOn 31 January 2025 the UK government published a voluntary code of 13 security principles for AI systems, intended as the basis of an ETSI global standard.
  2. IncidentData exposureDeepSeek left ClickHouse databases holding chat logs open to the internetIn January 2025 DeepSeek exposed two unauthenticated ClickHouse databases holding over a million log entries, including user chat history and API keys.
  3. Regulatory milestoneInternationalInternational AI Safety Report 2025 published ahead of Paris AI Action SummitThe first full International AI Safety Report, chaired by Yoshua Bengio, was published on 29 January 2025 ahead of the Paris AI Action Summit.
  4. Regulatory milestoneUS FederalExecutive Order 14179, Removing Barriers to American Leadership in AI, signedEO 14179, signed 23 January 2025, set a policy of US AI dominance, ordered an AI Action Plan within 180 days and a review of actions taken under EO 14110.
  5. Regulatory milestoneUS FederalExecutive Order 14148 revokes the 2023 AI order, EO 14110Signed 20 January 2025, Executive Order 14148 revoked EO 14110 on AI along with dozens of other orders and told officials to review actions taken under them.
  6. Regulatory milestoneUS FederalExecutive Order 14144 on cybersecurity signed, with new AI security provisionsPresident Biden signed EO 14144 on January 16, 2025, adding AI pilot programs and supply chain security rules for federal software.
  7. Regulatory milestoneUS FederalNIST releases second draft of AI 800-1 misuse risk guidance for foundation modelsUS AI Safety Institute published a second public draft of NIST AI 800-1 on managing misuse risk for dual-use foundation models on January 15, 2025.
  8. Regulatory milestoneUS FederalCISA and JCDC publish the AI Cybersecurity Collaboration PlaybookCISA's Joint Cyber Defense Collaborative releases a voluntary playbook for sharing AI-related cybersecurity incidents and vulnerabilities.
  9. Regulatory milestoneUKUK publishes the AI Opportunities Action Plan with 50 AI recommendationsStarmer government endorses all 50 recommendations of the AI Opportunities Action Plan, covering compute, data, skills and AI regulation.
  10. IncidentModel abuseGoogle discloses state-backed adversarial misuse of GeminiGTIG's January 2025 report details how Iranian, Chinese, North Korean and Russian state-linked actors tried to misuse Gemini, mostly without success.