第35期

AI审计周报 第35期

| ~

编者按：AI审计周报每周一发布，精选AI在审计、合规与监察领域的最新动态。中英文资讯混编，帮助审计人追踪行业前沿。

本期摘要

本期共收录 74 条与AI审计、合规和风险管理相关资讯，已按主题分组整理，便于快速浏览。

合规治理

When a Governance Lead Leaves, the Escalation History Often Goes With Them

  • 来源: Corporate Compliance Insights
  • 摘要: A handover judged only by whether meetings, papers and action logs stay current can let a long-running concern slide back to an earlier stage of scrutiny The post When a Governance Lead Leaves, the Es…
  • 标签: 合规治理

Marching Orders for DOJ’s Reshuffled Corporate Fraud Enforcers

  • 来源: Corporate Compliance Insights
  • 摘要: The DOJ’s newest division has given its corporate fraud prosecutors their most substantive public guidance yet in the form of a memo that outlines factors they are to consider when deciding whether to…
  • 标签: 合规治理

Smooth Talk: Why Organizations Mistake Communication for Competence

  • 来源: Corporate Compliance Insights
  • 摘要: When fluency & confidence pass for expertise, companies risk elevating the convincing over the capable The post Smooth Talk: Why Organizations Mistake Communication for Competence appeared first on Co…
  • 标签: 合规治理

Exploring California’s Generative AI Training Data Transparency Act

  • 来源: Corporate Compliance Insights
  • 摘要: AB 2013 obligations are live now, and the practical question is not whether to comply but how much to disclose The post Exploring California’s Generative AI Training Data Transparency Act appeared fir…
  • 标签: 合规治理

AI Meeting Assistants Raise Eavesdropping Concerns & Discovery Headaches

  • 来源: Corporate Compliance Insights
  • 摘要: A federal court has allowed wiretap and privacy claims against an AI notetaker to proceed The post AI Meeting Assistants Raise Eavesdropping Concerns & Discovery Headaches appeared first on Corporate …
  • 标签: 合规治理

AI行业

Western open-weight campaign could give enterprises more control over AI

  • 来源: AI Business
  • 摘要: New Western open-weight models could give enterprises an alternative to Chinese and proprietary AI, but greater control also brings new responsibilities…
  • 标签: AI行业

Anthropic opens cybersecurity program applications, releases cheaper Haiku

  • 来源: AI Business
  • 摘要: Under the program, the vendor provides enterprises with models that can defend against bad actors…
  • 标签: AI行业

Reflection Beam aims U.S. open-weight AI model at Chinese rivals

  • 来源: AI Business
  • 摘要: Nvidia-backed open source startup makes splashy entrance. AI hits U.S. K-8 education…
  • 标签: AI行业

The context factor for AI agents

  • 来源: AI Business
  • 摘要: Process Intelligence helps organizations provide AI agents with the right context and data needed to know how their businesses run…
  • 标签: AI行业

AI teaching systems raise new questions for school districts

  • 来源: AI Business
  • 摘要: As AI moves into K-8 instruction, district technology leaders must weigh external evidence, data protection and results-based contracts before scaling these systems…
  • 标签: AI行业

科技前沿

The Download: AI’s refusal problem and weight-loss drug side effects

  • 来源: MIT Technology Review
  • 摘要: This is today’s edition of The Download, our weekday newsletter that provides a daily dose of what’s going on in the world of technology. We’re putting too much faith in AI’s ability to say no Today’s…
  • 标签: 科技前沿

We’re putting too much faith in AI’s ability to say no

  • 来源: MIT Technology Review
  • 摘要: Ever since people first seriously contemplated giving machines an intelligence modeled on our own, there has never been any question that they would, like us, be able to say no. The sci-fi canon is fu…
  • 标签: 科技前沿

Job titles of the future: Delivery drone air traffic controller

  • 来源: MIT Technology Review
  • 摘要: The moment Trevor Wischnewsky heard that drones were delivering pizza and sushi around his Texas neighborhood, his mind was made up. “It was super fascinating,” he says. “I just immediately wanted to …
  • 标签: 科技前沿

Roundtables: A Conversation With the Creator of AI-Designed Viruses

  • 来源: MIT Technology Review
  • 摘要: Friday, October 16, 2026 Can AI design new life forms? In 2025, Stanford University PhD student Samuel King came up with a preliminary answer when he used a generative AI model to propose genetic blue…
  • 标签: 科技前沿

The Download: AI roadblocks for humanoids and portable rubber dams

  • 来源: MIT Technology Review
  • 摘要: This is today’s edition of The Download, our weekday newsletter that provides a daily dose of what’s going on in the world of technology. AI breakthroughs in robotics won’t change your life any time s…
  • 标签: 科技前沿

安全合规

CyberXero Uses Claude Code in Campaign That Stole Over 628,000 Ukrainian Records

  • 来源: The Hacker News
  • 摘要: Cybersecurity researchers have disclosed details of a two-pronged artificial intelligence (AI)-augmented initial access campaign targeting WordPress sites, e-commerce platforms, and critical infrastru…
  • 标签: 安全合规

P7 DarkSword iOS Exploit Kit Adds Crypto Wallet Data Theft and Remote Commands

  • 来源: The Hacker News
  • 摘要: Cybersecurity researchers have disclosed details of a previously unseen variant of the DarkSword iOS exploit kit called P7 DarkSword.

“Compared with the variants we usually observe, P7 reduces its on…

  • 标签: 安全合规

The Third-Party Agent Problem: Why Security Built for AI You Chose Misses the Agents You Didn’t

  • 来源: The Hacker News
  • 摘要: In environments studied for the 2026 State of Agent Security Report, roughly 1,280 third-party products now embed AI. About 282 of them sit behind single sign-on. The other thousand are invisible to i…
  • 标签: 安全合规

Anthropic Cuts Live Internet Access for Internal AI Tests After Claude Exploits Injection Flaws

  • 来源: The Hacker News
  • 摘要: Anthropic on Friday said it’s cutting off live internet access for all its internal evaluations until further notice following the discovery of new incidents in which its artificial intelligence (AI) …
  • 标签: 安全合规

Credential-Stealing GitHub Actions Workflows Planted in Tens of Thousands of Repositories

  • 来源: The Hacker News
  • 摘要: Cybersecurity researchers have disclosed details of an ongoing credential-theft campaign that has compromised two high-profile open-source maintainer accounts to push a malicious workflow into over 34…
  • 标签: 安全合规

合规科技

Structuring: the oldest trick, still the hardest to prove

  • 来源: Unit21 Blog
  • 摘要: Discover how Unit21’s AI Rule Writer and Investigation agents help compliance teams detect and prove structuring, turning complex cash splitting patterns into clear examiner-ready findings…
  • 标签: 合规科技

Corelation KeyStone integration: act on fraud from the case

  • 来源: Unit21 Blog
  • 摘要: See how the Unit21-Corelation KeyStone integration works: member data flows in for fraud and BSA/AML, and analysts block cards or stop payments from the case…
  • 标签: 合规科技

Alkami + Unit21: real-time fraud detection and AML in one platform

  • 来源: Unit21 Blog
  • 摘要: Unit21 is an Alkami Alliance Partner, giving Alkami financial institutions real-time fraud detection plus AML, case management, and FinCEN filing in one platform…
  • 标签: 合规科技

Progressive autonomy: the trust ladder for AI in compliance

  • 来源: Unit21 Blog
  • 摘要: Progressive autonomy is a five-level trust ladder for AI in compliance, from no AI (L1) to auto-closing false positives (L5). Includes a video walkthrough…
  • 标签: 合规科技

企业AI战略

For CPG companies, agentic AI’s real payoff is growth

  • 来源: McKinsey Insights
  • 摘要: As the relationship between consumer-packaged-goods companies and retailers becomes more complex, agentic AI can recapture lost sales. But are leaders ready to harness the technology’s power?…
  • 标签: 企业AI战略

Why AI’s easiest wins are misleading CEOs

Author Talks: Rana Foroohar on a new ‘great game’ in the Arctic

  • 来源: McKinsey Insights
  • 摘要: From shipping and rare earths to energy and security, the Arctic is gaining importance in the global economy. The Financial Times’ Rana Foroohar explores its role in a new era of great-power competiti…
  • 标签: 企业AI战略

When Compliance Workarounds Backfire

  • 来源: MIT Sloan Management Review
  • 摘要: John Holcroft/Ikon Images In 2021, Barcelona-based delivery platform Glovo was faced with a law that threatened its business model. Spain had passed the Rider Law, which introduced a presumption of em…
  • 标签: 企业AI战略

The Real Reasons Why Digital Projects Fail

  • 来源: MIT Sloan Management Review
  • 摘要: Danae Diaz/Ikon Images It’s a familiar scenario: An organization implements digital tools intended to improve a particular process but fails to obtain the desired result. Employees who are meant to us…
  • 标签: 企业AI战略

企业AI应用

Governed context: The key to scaling enterprise AI

  • 来源: CIO.com
  • 摘要: The AI revolution is experiencing some growing pains. Across industries, business leaders are confronting the same challenge: AI pilots dazzle, but they don’t scale. An agent that reasons brilliantly …
  • 标签: 企业AI应用

Why CIOs should choose Samsung Galaxy Enterprise Edition

  • 来源: CIO.com
  • 摘要: CIOs, it’s time to rethink your mobile device lifecycle Let’s start with an uncomfortable question: how much risk is hiding in devices that are no longer receiving security updates? Enterprise mobilit…
  • 标签: 企业AI应用

AI won’t empty the software factory: Why the new engineers act like foremen

  • 来源: CIO.com
  • 摘要: Human work has gone through several revolutions over the past few decades, each of them bringing a start to new careers and an end to others. Software careers, which seemed virtually untouchable a sho…
  • 标签: 企业AI应用

Your AI foundation must extend without surrendering control

  • 来源: CIO.com
  • 摘要: AI models are becoming short-term tenants in long-lived enterprise data estates. Yet many organizations are remodeling the building around each new occupant. The market keeps asking which model enterp…
  • 标签: 企业AI应用

The real agent risk that Replit’s database deletion revealed

  • 来源: CIO.com
  • 摘要: In July 2025, Jason Lemkin, founder of SaaStr and a well-known investor in enterprise software, decided to publicly test one of AI’s most repeated promises: that anyone can build an application withou…
  • 标签: 企业AI应用

AI科技

Taking a look under your agent’s hood

  • 来源: Stack Overflow Blog
  • 摘要: Ryan is joined by Yanbing Li, Chief Product Officer at Datadog, to talk about applying observability to non-deterministic AI agents, blurring the boundaries between software development and production…
  • 标签: AI科技

Production-grade LLMs and agents: a field guide

  • 来源: Stack Overflow Blog
  • 摘要: A maturity model for taking agents from an impressive demo to a system people can depend on — with a self-assessment and the map to a deep-dive on each layer…
  • 标签: AI科技

Part 5: Operating an LLM system: observability, cost, routing, and the platform underneath

  • 来源: Stack Overflow Blog
  • 摘要: Your service can be 100% up and still quietly approving the wrong things, burning its budget, or failing over into untested quality. Level 5 is the infrastructure that lets you see your decisions, bou…
  • 标签: AI科技

Part 4: Safety and governance for LLM systems: guardrails, PII, audit, and memory

Part 1: Make your AI agents boring: the determinism layer

AI前沿

Unlocking Earth AI’s planetary geospatial foundation models for global public health

Open and Emergent Problems in Agentic Privacy and Security: A Contextual Angle

#AI审计 #合规科技 #内部审计