Business · 17 February 2026

AI in Business

For business leaders and professionals

At askKira, we provide secure, personalised enterprise AI for UK businesses. Our platform prevents data leaks, protects IP and client data, and gives your teams AI-powered tools at a fraction of the cost of other enterprise solutions.

This week's roundup brings you the latest AI developments that matter for UK organisations.

Legal & Regulatory Landscape

UK regulators unite on AI financial conduct

The Financial Conduct Authority (FCA), ICO and Bank of England have strengthened their collaboration through the Digital Regulation Cooperation Forum. This joint approach clarifies how existing financial conduct rules apply to AI, specifically addressing conflicts between data protection and financial regulations. The bodies are prioritising a technology-neutral framework that holds senior leadership accountable for algorithmic decisions under the Senior Managers and Certification Regime.

Why it matters: Financial services firms can no longer treat AI compliance as a siloed IT issue; governance must now bridge data protection and financial conduct to satisfy multiple regulators simultaneously.

Read the source on LinkedIn →

Legal privilege denied for AI-generated documents

In a significant ruling with implications for global legal teams, a US federal judge has determined that AI-generated documents do not automatically qualify for attorney-client privilege simply because they were sent to a lawyer. Judge Jed Rakoff ruled that privilege protects communications for legal advice, but if an executive uses AI to draft a narrative or summary independently, that material may be discoverable in court.

Why it matters: UK legal teams must urgently review internal protocols, as executives using AI to 'prep' for legal counsel may be inadvertently creating discoverable evidence that is not protected by privilege.

Read the source on LinkedIn →

UK moves to criminalise non-consensual deepfakes

New legal provisions are coming into force that explicitly target the creation and distribution of non-consensual deepfake content. This legislative update closes loopholes regarding synthetic media, treating the generation of such content with similar severity to other forms of image-based abuse. This move places a higher burden on platforms and employers to monitor the use of generative tools within their networks.

Why it matters: HR and safeguarding policies must be updated immediately to classify the creation of deepfakes using company hardware as gross misconduct with potential criminal liability.

Read the source on LinkedIn →

Enterprise Security & Strategy

Security blind spots found in multi-LLM agents

A new study reveals a critical security gap when DevOps teams stack AI agents from different providers. Security tools that are 93% accurate on one provider's agents dropped to 49% accuracy when monitoring others, creating a 'detection gap' that attackers can exploit. Traditional security tools relying on timing and signatures are failing to detect threats across diverse model behaviours.

Why it matters: If your organisation uses a 'bring your own model' approach, your current security stack may be failing to detect nearly half of all malicious injections or data exfiltration attempts.

Read the source on LinkedIn →

AI adoption driven by incentives, not age

New research from McKinsey dispels the myth that AI adoption is a generational issue. The study found that 'digital natives' adopt AI faster only because they are incentivised to learn (grades or career entry), while senior leaders often lack similar motivation. The bottleneck is not technical literacy but the absence of KPIs and incentives that encourage senior staff to alter established workflows.

Why it matters: Stop relying on 'reverse mentoring' from junior staff and start redesigning executive compensation and performance metrics to reward AI-first leadership behaviours.

Read the source on LinkedIn →

Agentic Trends

Non-technical staff driving 'Agentic Coding' boom

The Anthropic 2026 Agentic Coding report highlights that non-technical professionals are now producing 27% more output by building their own software solutions using AI agents. Staffing teams, operations managers and designers are bypassing traditional IT queues to build bespoke tools. While this boosts productivity, it raises significant concerns regarding code quality and 'shadow IT' governance.

Why it matters: IT leaders must urgently decide whether to block these tools or provide a sanctioned 'sandbox' environment, as unmanaged code generated by non-developers poses long-term maintenance risks.

Read the source on LinkedIn →

On your radar

Technical Risk OwnershipDo not let your technical teams hand off AI governance entirely to Legal or Risk departments. As noted in recent governance discussions, technical product owners must retain accountability for model drift and data provenance throughout the lifecycle, rather than viewing compliance as a one-time pre-deployment check.
The 'Freshly Hired' MindsetWhen deploying new AI agents, shift your internal messaging from 'software installation' to 'employee onboarding'. Treat the AI as a capable but inexperienced graduate who requires context, correction and supervision for the first month, rather than expecting perfection out of the box.
SaaS vs. Vibe CodeDespite the hype around AI replacing software, recent earnings calls from major tech firms suggest enterprise SaaS is here to stay. Avoid the trap of trying to replace mission-critical platforms (like Salesforce) with bespoke AI agents ('vibe code') that lack the security, audit trails and reliability of established enterprise software.
The Data GapBefore investing in agentic AI, assess your data readiness. Many organisations are attempting to run advanced reasoning models on static, low-context data (PDFs and old Intranets). If your data isn't structured for retrieval, even the most advanced model will fail to deliver ROI.
NewerAI in Business, 24 February 2026OlderAI in Business, 10 February 2026