Read time: under 9 minutes

This week, AI governance moved from principles to deadlines and bans. In Europe, the Council backed the Digital Omnibus plan to streamline AI Act implementation, set new dates for high-risk obligations, restored safeguards, and added an explicit ban on systems used to generate non-consensual sexual or intimate content and child sexual abuse material. In parallel, Brussels is advancing a separate ban on “nudification” apps after the Grok fallout.

In the U.S., the legal perimeter is tightening too. New York’s S7263, sold as an anti-impersonation bill, faces criticism for being broad enough to chill legitimate legal information even with clear AI disclosures. And in court, a judge ordered UnitedHealth to produce years of documents in an AI denial case, pulling incentives and internal governance into discovery. On the market side, Harvey is hedging toward an ecosystem, investing in tools it doesn’t plan to build.

Our feature this week tracks the same shift: Sandstone’s model for an AI-native legal department where knowledge is operational, not just searchable.

➡️ EU Council Agrees Position to Streamline AI Act Implementation | EU governments have agreed a Council position on the Commission’s “Digital Omnibus” plan to simplify implementation of the AI Act, including delaying high-risk AI obligations with fixed start dates of 2 December 2027 for stand-alone high-risk systems and 2 August 2028 for high-risk AI embedded in regulated products; the Council also adds an explicit ban on AI used to generate non-consensual sexual or intimate content and child sexual abuse material, restores certain safeguards, postpones national AI regulatory sandbox deadlines, and now moves into negotiations with the European Parliament.
^{Mar 13, 2026, Source: The Legal Wire}

➡️ NY Bill Targeting “AI Lawyer” Chatbots Criticized as Too Broad | A new critique argues New York’s S7263, pitched as an anti-impersonation bill, would create sweeping liability for chatbots that give “substantive” legal information, without requiring any fake credentials or even concealment that the tool is AI. Because disclaimers wouldn’t shield providers and key terms aren’t defined, critics warn it could chill legitimate legal info tools (including lawyer-supervised use) and invite opportunistic lawsuits. The piece urges lawmakers to rewrite it as a narrower ban on clearly deceptive claims of licensure.
^{Mar 13, 2026, Source: Bloomberg Law}

➡️ Harvey To Invest in Legal Tech Startups, Betting on the Next Wave | Harvey is moving into startup investing through a partnership with The LegalTech Fund, to back niche tools it doesn’t plan to build itself. CEO Winston Weinberg says the legal market is too large and fragmented for one platform to dominate, so Harvey will use its own revenue, writing sub-$2M checks, to support emerging vendors in areas like intake or patent workflows. Some investments could become integrations or acquisitions, with security standards as the gate. The move mirrors a broader trend of major AI startups launching investment arms, and reflects capital increasingly shaping legal tech’s consolidation and partnership landscape.
^{Mar 11, 2026, Source: Business Insider}

➡️ EU Moves to Ban AI “Nudification” Apps After Grok Deepfake Fallout | EU lawmakers and member states are advancing proposals to ban AI systems marketed in Europe that can generate non-consensual sexualized images, videos, or audio of real people. The push follows backlash over X’s Grok, which was used to create large volumes of sexualized deepfakes, including of minors, prompting an ongoing EU probe into whether X mitigated risks under the Digital Services Act. The ban could take effect as early as this summer, pending final agreement between the Council and Parliament, though drafts suggest potential carve-outs if providers can prove effective safeguards that prevent generation and misuse.
^{Mar 11, 2026, Source: MLex}

➡️ Judge Orders UnitedHealth to Produce Documents in AI Denial Lawsuit | A federal magistrate judge in Minnesota ordered UnitedHealth Group to turn over extensive discovery in a case alleging its Medicare Advantage plans used the nH Predict tool (Optum/naviHealth) to wrongfully deny or shorten post-acute care. The court required policies and internal analyses dating back to 2017, materials tied to the naviHealth acquisition and cost savings, records of government scrutiny, and details on staff incentives and an internal AI review board, plus names of personnel involved in denials for 300 proposed class members. The judge declined to compel source code and some broad financial/HR requests. UnitedHealth has 21 days to comply.
^{Mar 10, 2026, Source: Law360}

If you listen closely to how in-house legal teams talk about their work, you’ll notice a pattern emerging fairly quickly. The frustration isn’t usually about the law itself, but everything around it. The constant Slack messages, the repeated questions, the same clauses being rewritten, reviewed, and negotiated again and again. And the lingering sense that a lot of institutional knowledge exists, but never quite shows up when it’s needed.

For years, legal technology has tried to solve this with more tools. New dashboards, new AI assistants, and new ways to search documents faster. Some of those tools help, but many don’t stick. And most struggle with the same underlying problem: legal work doesn’t happen in isolation, and neither does legal knowledge.

That’s the context in which Sandstone has entered the conversation. In January this year, Sandstone announced a $10 million seed round led by Sequoia, and described itself as “the platform for AI-native legal departments.” It’s a bold phrase, and an intentionally structural one. Rather than promising faster answers or better drafts, Sandstone is making a different claim: that the real opportunity lies in turning institutional legal knowledge into something operational.

Not stored. Not searched. Used.

Click here to continue reading the full interview

The AI Regulation Tracker offers a clickable global map that gives you instant snapshots of how each country is handling AI laws, along with the most recent policy developments.

The most recent developments from the past week:

📋 14 March 2026 | US Commerce Department withdraws planned rule on AI chip exports: The US Commerce Department has withdrawn a planned rule on AI chip exports, marking a reversal by the Trump administration in its strategy to secure American leadership in AI. The draft rule, intended to replace a Biden-era regulation on global access to AI chips, was circulated for feedback but was ultimately pulled, with officials stating it was always a preliminary draft. The proposed rule had suggested requiring foreign investments in US data centers or security guarantees for exporting large quantities of chips, contrasting sharply with the Biden administration's tiered approach to chip distribution based on geopolitical considerations.

📋 12 March 2026 | Indonesia sets rules for AI, digital tech use in education: The Indonesian government has issued a joint ministerial decree regulating the use of digital technology and AI in education, encompassing all levels from early childhood to higher education. Signed by seven cabinet ministers, the decree establishes guidelines on the minimum age for technology use, permissible types of use, and recommended durations, tailored to different educational stages. For younger students, particularly in early childhood and primary education, the regulations focus on controlling screen time and content. The decree restricts the use of instant AI applications that generate direct answers for primary and secondary students, allowing only AI tools specifically developed for educational purposes.

📋 11 March 2026 | Safety bodies, experts warn against softening EU AI Act's overlap with sectoral rules: It is reported that a coalition of AI safety organizations and experts, including participants in the AI Act's standardization process, have expressed concern over a European Parliament proposal within the AI omnibus amendment package that aims to relax the AI Act's requirements for products already governed by sector-specific safety regulations. They argue that such a softening could undermine the effectiveness of the AI Act and compromise safety standards.

📋 11 March 2026 | China moves to curb use of OpenClaw AI at banks, state agencies: It is reported that the Chinese government has issued directives to state-run enterprises and government agencies, including major banks, advising against the installation of OpenClaw AI software on office devices due to security concerns. Employees who have already installed the software are instructed to inform their superiors for security assessments and potential removal. OpenClaw, an open-source AI agent capable of autonomously performing tasks such as file sorting and internet browsing, has raised alarms over its extensive access to private data and potential vulnerability to external attacks.

🆕 Sirion - Conversational AI for enterprise contracting. Effortless conversations meet purpose-built AI agents. Sirion agentic CLM powers contracting with enterprise-grade precision and trust at every step.

🆕 Vesence - AI in Microsoft Office for law firms and professional services.

🆕 Ruli AI - Continuous legal intelligence. Your legal team has a memory, Ruli helps it speak.

Want more Legal AI Tools? Check out our
Top AI Tools for Legal Professionals

Why it helps: Gives you a ready-to-adopt AI governance baseline, so you can use AI confidently, reduce risk, and align the whole team on what’s allowed.

The Responsible AI Collaborative is a not‑for‑profit organization working to present real‑world AI harms through its Artificial Intelligence Incident Database.

View the latest reported incidents below:

⚠️ 2026-02-23 | Anthropic Said DeepSeek, Moonshot, and MiniMax Used Fraudulent Accounts and Proxies to Illicitly Distill Claude Capabilities at Scale | View Incident

⚠️ 2026-01-28 | South Korean Woman Allegedly Used ChatGPT to Assess Lethality of Drug-and-Alcohol Mixtures Before Two Fatal Motel Poisonings | View Incident

⚠️ 2026-01-21 | DHS Agents Reportedly Threatened Legal Observers With 'Domestic Terrorist' Database While Using Purportedly AI-Enabled Surveillance During ICE Operations | View Incident

The Legal Wire is an official media partner of:

Thank you so much for reading The Legal Wire newsletter!

If this email gets into your “Promotions” or "Spam” folder, move it to the primary folder so you do not miss out on the next Legal Wire :)

If you have any tips for us, just reply to this e-mail! We’d love any feedback or responses from our readers 😄

The Legal Wire takes all necessary precautions to ensure that the materials, information, and documents on its website, including but not limited to articles, newsletters, reports, and blogs ("Materials"), are accurate and complete.

Nevertheless, these Materials are intended solely for general informational purposes and do not constitute legal advice. They may not necessarily reflect the current laws or regulations.

The Materials should not be interpreted as legal advice on any specific matter. Furthermore, the content and interpretation of the Materials and the laws discussed within are subject to change.