In this week’s session, Tomislav Pedišić, Managing Partner of VUKMIR, explored how ITech Law’s European meeting in London gathered more than 500 practitioners across 70+ jurisdictions, reflecting the profession’s appetite for face-to-face exchange and the resurgence of tech-focused mandates. Pedišić, an M&A and technology specialist, reported on two tracks ( namely Interactive Entertainment & Media and Tech M&A), drawing out what matters now for dealmakers, advisors, and in-house teams.

Regulation vs Innovation: Navigating the Platform Rules

He spoke about the European Digital Services Act (and the UK’s counterpart), which deals with a through-line: content moderation, advertising transparency, and protection of minors, which now sit at the centre of platform compliance. Pedišić’s take was that balanced systemic risks justify guardrails, but the burden on startups and scaleups is real. The profession’s task is to make compliance tractable, especially for smaller teams with our limited runway of cash.

NIS2 and Patchy Implementation

Cybersecurity obligations under NIS2 are unevenly implemented across the EU. Some member states (including Croatia) have moved faster than others, leaving multinationals to operate in a fragmented landscape. The bottleneck is often a result of limited state capacity rather than resistance, so counsel should plan for rolling localisation.

Tech M&A: What Changes When AI Enters the Room

Within the Tech M&A committee, practitioners agreed to publish model representations and warranties tailored to AI-involved targets a practical step toward common standards. Diligence is shifting from generic questionnaires to AI-specific scrutiny: data provenance and licensing; lawful bases for processing; model governance and evaluation; subcontractor IP and confidentiality; and product security posture.

Watch-outs for in AI deals:

• Data lineage & rights: Where training and fine-tuning data originated; auditability and deletion rights.
• Third-party code & models: Open-source licences, API dependencies, and model-as-a-service terms.
• Safety & performance claims: Evidence behind marketing statements; evaluation datasets; drift monitoring.
• Customer promises: SLAs and warranties that silently allocate model risk back to the seller.

Law-Firm AI: Useful, But Mind the Guardrails

Firms are experimenting—marketing copy, first-pass drafting, and legal research with tools like Harvey and Legora—yet language coverage and data-security boundaries limit broader deployment. Pedišić emphasised that internal policies (no sensitive client data in public tools; mandatory transparency to supervising lawyers when AI is used) need to be drafted and consistently monitored.

Disputes Go Digital: AI in Arbitration (Pilot)

Pedišić pointed out a notable development the American Arbitration Association is piloting AI-supported arbitration for construction disputes up to USD 100,000. AI will generate a draft summary of facts and a proposed decision for the arbitrator; the parties can review and comment before a human-signed award issues. Early use cases are “commodity-type” matters with high document regularity, promising speed and cost without displacing adjudicative judgment.

Cookies After the Cookie Banner: PETs on the Rise

With the long-mooted e-Privacy Regulation stalled, the current directive still requires consent for non-essential cookies. Privacy-enhancing technologies (PETs)—including anonymisation and pseudonymisation techniques may reduce dependency on third-party cookies and the associated consent friction, but adoption depends on coordinated action between brands and ad-tech vendors.

What General Counsels and Deal Teams Should Do Next

1. Operationalise AI diligence: Add data provenance and model-risk annexes to your standard diligence packs.
2. Localise cybersecurity compliance: Stand up jurisdiction-specific NIS2 playbooks.
3. Refresh platform governance: Align trust-and-safety, ads transparency and minor-protection policies with DSA/UK norms.
4. Codify firm-side AI use: Policy, training, and disclosure keep it simple and auditable.
5. Scan disputes portfolios: Where claims are repeatable and low-value, explore AI-assisted procedures to compress cycle time.

Bottom line: The profession is moving from abstract “innovation talk” to concrete artefacts, model clauses, playbooks, and pilots that accelerate execution while containing risk. Lawyers who package complexity into repeatable workflows will set the pace in 2026, and we look forward to ITech law in the coming year and further practice updates.