Read time: under 4 minutes

Welcome to this week's edition of The Legal Wire!

A global contract-drafting study finds AI now neck-and-neck with humans: Gemini 2.5 Pro slightly beat the top lawyer on reliability, and specialist tools flagged risks people missed. The big edge wasn’t raw drafting but workflow, intake, redlining, and review. Nearly all lawyers use AI, but only ~35% use legal-specific platforms.

Governments and firms are formalizing the shift. The U.S. cleared Meta’s Llama for federal pilots; Ireland named 15 AI regulators and set up a single contact point ahead of the EU AI Act. In the UK, most top firms run third-party AI and many are building their own, leaving smaller shops at risk of falling behind.

Reality check: a California appeals court fined an attorney $10,000 after 21 of 23 citations in his brief were ChatGPT fabrications, a warning that unverified AI outputs won’t fly in court.

And Josef turns repeatable in-house requests into self-service tools (answers, contracts, approvals) so legal becomes an enabler, not a bottleneck. Read the full interview below.

This week’s Highlights:

• Industry News and Updates

• From Blockers to Enablers: How Josef Helps In-house Legal Teams Drive the Business

• AI Regulation Updates

• AI Tools to Supercharge your productivity

• Legal prompt of the week

• Latest AI Incidents & Legal Tech Map

➡️ California Lawyer Fined $10K for ChatGPT-Cited Fake Cases | Cal Matters reports: A California appeals court fined attorney Amir Mostafavi $10,000 after 21 of 23 case quotes in his brief were found to be fabricated by ChatGPT. The court warned lawyers against filing unverified citations as state authorities scramble to regulate AI in the judiciary. Mostafavi admitted he failed to check the AI-generated text, saying, “I’m paying the price.” Experts warn such incidents will rise as more lawyers adopt AI tools prone to fabricating cases.
^{Sep 22, 2025, Source: Cal Matters}

➡️ AI vs. Lawyers in Contract Drafting: New Global Study | Over 500 experts across 34 countries tested how AI tools stack up against human lawyers in contract drafting. The results show AI is catching up fast. Google’s Gemini 2.5 Pro slightly outperformed the best human lawyer in reliability (73.3% vs. 70%), while specialized legal AI tools flagged risks lawyers missed. Still, workflow support, not just raw drafting quality, proved the biggest advantage of legal AI platforms. Key takeaways: 97% of lawyers now use AI for legal work, but only 35% use legal-specific tools. ChatGPT leads adoption (76%), while platforms like August, Brackets, and GC AI scored high in usefulness and workflow support.
^{Sep 23, 2025, Source: Legalbenchmarks.ai}

➡️ Meta’s Llama AI Approved for U.S. Government Use | The U.S. General Services Administration has approved Meta’s AI system Llama for use across federal agencies, allowing them to experiment with the free large language model under government security and legal standards. Llama joins a roster of AI tools from Amazon, Microsoft, Google, Anthropic, and OpenAI, which have also been cleared, often at steep discounts. Officials say the move will let agencies test AI in areas such as contract review and IT troubleshooting. GSA procurement lead Josh Gruenbaum emphasized the partnerships are about strengthening national capacity, not political favor.
^{Sep 22, 2025, Source: Reuters}

➡️ Top Law Firms Embrace AI to Stay Ahead | Three-quarters of the UK’s 20 largest law firms now use third-party AI tools, while nearly half have built or customized their own, according to new Thomson Reuters research. The majority also train staff in AI skills, invest in startups, or run incubators, while all but one publicly promote their AI use. Many are creating new divisions to advise clients on AI, with a third publishing ethics frameworks. Smaller firms are lagging behind, raising concerns about a widening gap. Meanwhile, CILEX has launched an AI Academy to help legal professionals upskill quickly and keep pace with rapid developments.
^{Sep 22, 2025, Source: LegalFutures}

➡️ Ireland Names 15 AI Regulators, Sets Up National AI Office | Ireland has taken major steps toward implementing the EU AI Act, announcing 15 national competent authorities to oversee compliance, a single point of contact at the Department of Enterprise, Trade and Employment, and plans for a National AI Office by August 2026. The new office will coordinate regulators, act as Ireland’s main EU contact, and host a regulatory sandbox to drive AI innovation and compliance. Officials, including Enterprise Minister Peter Burke and AI Minister Niamh Smyth, said the move provides clarity for businesses. Legal experts expect more draft legislation and resourcing details before the office is fully operational.
^{Sep 18, 2025, Source: Law Society Ireland}

Legal Technology

From Blockers to Enablers: How Josef Helps In-house Legal Teams Drive the Business

Somewhere between your fourth “Can I do this?” Slack message of the day and your tenth search through a dusty SharePoint folder, your brain returns to that existential question: is this what you went to law school for?

The team at Josef doesn’t think so. That’s why they built a platform designed to help legal and compliance teams get unstuck, and help the business keep moving.

Founded with the mission to help legal and compliance teams actually get things done, Josef is an AI-powered platform where in-house teams at the likes of L’Oréal, Bumble and Randstad create self-service tools to handle the repeatable work that slows everyone down. Whether it’s answering questions, generating contracts, or managing intake and approvals, Josef puts the right information and processes in the hands of the people who need them, when they need them.

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:

📋 19 September 2025 | Brazil enacts sweeping bill requiring online age verification, safeguards for children’s data: Brazilian President Luiz Inácio Lula da Silva has signed a new law (LAW No. 15,211) requiring digital service providers to verify users’ ages and implement stringent data protection measures for children and adolescents. The legislation mandates that tech companies employ reliable age verification mechanisms to prevent minors from accessing content involving violence, pornography, sexual exploitation, drugs, gambling, or self-harm. Additionally, platforms must establish parental supervision tools, prohibit the processing of children’s personal data in ways that infringe on their privacy, and ban the use of such data for targeted advertising. This law, which updates a 1990 statute, is set to take effect in March, positioning Brazil as the first Latin American country to enact dedicated online privacy and safety protections for children.

📋 17 September 2025 | Italy finally approves AI bill, covering privacy, oversight and child access: Italy’s parliament has approved a new law (Senate Act No. 1146-B) covering AI, making it the first EU country with comprehensive AI regulations aligned with the EU AI Act. Key features include: (1) prison sentences of between one and five years for the illegal spreading of AI-generated or manipulated content if it causes harm; (2) harsher penalties for using AI to commit crimes, including fraud and identity theft, and stricter transparency and human oversight rules governing how AI is used in workplaces as well as in a range of sectors such as healthcare, education, justice and sport; (3) Children under the age of 14 will need parental consent to access AI; (4) On copyright, works created with AI assistance are protected if they originate from genuine intellectual effort, while AI-driven text and data mining will only be permitted for non-copyrighted content or scientific research by authorised institutions; (5) the Agency for Digital Italy and the National Cybersecurity Agency have been appointed to enforce the legislation; and (6) the law authorises up to €1bn (£870m) from a state-backed venture capital fund to support companies active in AI, cybersecurity and telecommunications.

📋 16 September 2025 | Vietnam plans to introduce Law on Artificial Intelligence in 2025: It is reported that Vietnam plans to enact an AI Law by the end of 2025 to establish a comprehensive legal framework for AI, as announced by Minister of Science and Technology Nguyen Manh Hung at a national scientific conference. The law aims to position AI as the country's intellectual infrastructure, serving the populace, promoting sustainable development, and enhancing national competitiveness. Additionally, Vietnam intends to establish a national AI supercomputing center and a shared open AI data platform, universalize AI literacy, and develop a national AI Ethics Code aligned with international standards but tailored to the country's context. The government also plans to increase spending on AI, with the National Technology Innovation Fund allocating at least 40% of its resources to support AI adoption, including issuing vouchers for small and medium-sized enterprises to use Vietnamese AI solutions.

🆕 Smith.ai - Conversations designed to move your business forward, with AI precision and human expertise.

🆕 LegalReview - Get insights quickly, identify important terms, and analyse changes, before sending to lawyers for final review.

🆕 Diligen - Identify key provisions, generate contract summaries and help your team manage review with machine learning powered analysis.

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

This prompt eliminates hours of hunting and stitching instructions, delivering ready drafts, a verdict form roadmap, and preservation notes so you can focus on evidence and argument.

Instructions:
Provide your jurisdiction, causes of action/defenses, and any special issues (comparative fault, punitive damages, agency, etc.). Ask ChatGPT to:

1. List all applicable pattern jury instructions with citations and brief one-line summaries.

2. Generate a tailored draft of each instruction, customized to your facts (neutral, court-ready language).

3. Build a verdict form outline (questions in logical order, yes/no flow).

4. Flag preservation points (objections, alternative language) and common reversible-error traps.

5. Provide a checklist of elements per claim/defense to confirm proof at trial.

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:

⚠️ 2025-09-07 | Russian Disinformation Campaign Reportedly Used AI-Generated Posts and Videos to Target 2025 Moldovan Parliamentary Elections | View Incident

⚠️ 2025-08-27 | Anthropic Reportedly Identifies AI Misuse in Extortion Campaigns, North Korean IT Schemes, and Ransomware Sales | View Incident

⚠️ 2025-08-28 | Meta AI on Instagram Reportedly Facilitated Suicide and Eating Disorder Roleplay with Teen Accounts | 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 :)

Did we miss something or do you have tips?

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

Disclaimer

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.