General Counsel have spent their careers developing legal judgement.
Within their organisations, they may also have spent years shaping legal strategies, establishing risk positions, understanding the history behind key decisions, and working out how to protect the organisation while still helping the business achieve its commercial objectives.
That judgement stems from understanding the organisation, its appetite for risk, previous advice, regulatory relationships, commercial priorities and the consequences of getting a decision wrong.
Then a commercial manager walks in with a 12-page AI-generated presentation describing a particular legal issue they think they’ve discovered, the applicable law and what the company should do next. Or a senior executive, having spent half an hour interrogating an AI tool, starts explaining the legal position to the General Counsel.
For many senior in-house lawyers, this is becoming an uncomfortable part of working with generative AI.
It is frustrating to have carefully developed legal strategies challenged by someone whose confidence is built on an AI response generated without the same context, experience, or accountability. It can also be personally difficult. Senior lawyers are accustomed to having their advice discussed.
Still, there is a difference between constructive challenge and suddenly having your professional expertise treated as equivalent to an answer produced from a prompt. The challenge for General Counsel is how to respond without becoming defensive or positioning the legal department as being resistant to AI.
This is tricky, as they need to protect the professional role of General Counsel, the legal function, and the integrity of the organisation’s legal strategy, while also potentially dealing with a much broader problem: employees may be entering sensitive information into AI tools, generating new records, circulating legal analysis and potentially creating confidentiality, privilege and discovery issues about a “legal issue” that has already been carefully considered and addressed.
AI Has Changed the Dynamic Between Legal and the Business
Legal teams have spent years encouraging the business to self-service, within set guardrails.
Templates, playbooks, contract guidance, intake forms, FAQs and legal portals have all been designed to help the business resolve straightforward issues without waiting for a lawyer. But those tools have traditionally operated within boundaries set by the legal department. AI changes that.
An employee can now identify what they think the legal problem is, select the facts they think are relevant, ask the AI to interpret them, test alternative arguments and develop a recommended legal position and Legal may not be involved at any stage.
That is a fundamentally different form of self-service and one to be prepared for. The problem is not that employees suddenly have access to legal information. They have always had access to Google and external resources. What has changed is the experience.
Generative AI does not simply provide information. It explains, synthesises, argues and recommends. It can produce something that looks remarkably similar to professional legal analysis.
That fluency can create confidence well beyond the user’s actual legal understanding.
The AI Answer Does Not Carry the Accountability
This is where the distinction between legal information and legal judgement matters. An AI tool may be able to explain a legal principle, and it may even identify and quote relevant legislation, contractual provisions or regulatory guidance. But what it does not understand is why the organisation adopted its current position.
It may not know that the legal team has previously negotiated a particular regulatory outcome. It may not understand the organisation’s contractual history, litigation exposure, insurance arrangements, internal delegations, risk appetite or previous advice from external counsel. Nor does it carry responsibility for the decision that follows. The General Counsel does. That is why this issue can be so professionally uncomfortable.
When a colleague says, “AI says we can do this”, the GC is not simply being asked to debate another source of information. They may be asked to abandon or defend a legal strategy developed with far greater context than the AI has been given.
Perhaps the answer is to be much clearer about who owns legal judgement, the role of the in-house legal function, and the risk and responsibilities that sit with the business.
General Counsel Should Not Have to Compete with AI
Trying to win an argument against every AI-generated legal opinion is unlikely to work.
It also puts the GC on the defensive. The conversation should not become my legal opinion versus your AI’s legal opinion. Instead, the General Counsel can reset the discussion around context, accountability and decision rights.
Questions such as these may be useful:
- What information was supplied to the AI?
- What assumptions did it make?
- Did it have access to our existing contracts, policies or previous legal advice?
- Was confidential or privileged information entered into the system?
- Which AI system was used, and is that approved for this type of information?
- Has the analysis been circulated or incorporated into other documents?
- What commercial outcome are we trying to achieve?
That changes the conversation.
It acknowledges that AI can contribute useful information while making clear that legal advice within an organisation is not determined by whichever person produces the most convincing AI response.
There May be a Governance Problem Hiding Behind the Frustration
The professional tension is only one part of the problem.
By the time an AI-generated legal analysis reaches the legal team, the organisation may already have created serious legal and information governance risks. An employee may have uploaded contracts, correspondence, personal information, internal investigations or previous legal advice into an AI system.
They may then have generated multiple prompts and responses, copied the output into a presentation, saved it to a shared folder, and circulated it to colleagues. Each of those actions is concerning.
Enterprise AI environments may offer stronger contractual, security and data-handling protections than publicly available tools. That can certainly reduce some risks, but it does not automatically answer questions about access controls, confidentiality, privacy, legal professional privilege, records management or discovery.
Legal professional privilege requires particular care. An employee’s prompt, the resulting AI response and documents subsequently created from that material should not be assumed to have the same privileged status as confidential lawyer-client advice.
Equally, using AI does not automatically mean privilege has been lost. The circumstances, purpose, information involved and way the material is handled all matter.
Discovery and records management raise another issue. The relevant material may no longer be limited to the final document. Prompts, uploaded material, AI responses, summaries, and presentations may all become information that needs to be identified, preserved, or reviewed, depending on the matter and the organisation’s obligations.
For General Counsel, this means the problem often begins before anyone formally asks for legal advice.
Draw a Line Between Exploration and Legal Advice
Organisations need a clearer boundary between using AI to explore an issue and using AI to investigate legal issues and legal positions. This means policies and individuals across the organisation are clear about when the legal department must become involved and the risks of not conforming to these policies.
A practical model could include three categories:
- Permitted self-service: using approved AI tools to organise non-sensitive information, locate approved internal guidance, prepare questions or complete structured legal intake using an approved legal intake system.
- Conditional use: using an approved AI environment to prepare an initial issue summary or explore possible considerations, subject to defined information controls and human review.
- Mandatory Legal involvement: matters involving potential legal exposure, disputes, investigations, regulatory issues, sensitive or potentially privileged information, significant contractual consequences, high-value decisions, novel legal questions or possible departures from existing legal positions.
These categories should be supported by examples that make sense to the organisation and the different personas across the business.
Protect The Legal Strategy Without Becoming the AI Police
There is also a leadership question for General Counsel. How do you protect the authority of Legal without appearing territorial? Part of the answer is to make explicit something that may previously have been assumed. AI can assist with legal work, but it does not determine the organisation’s legal position. That responsibility still sits with appropriately authorised people.
For some organisations, this may need to be reflected in AI policies, delegations, legal intake processes and governance standards. For others, it may require changing how Legal communicates its role.
It also gives General Counsel a stronger foundation when their advice is challenged. The conversation is no longer about defending professional status against technology. It entails ensuring that legal decisions are made with the right context, information, authority and accountability.
Make It Easier to Come to Legal Before They Reach For AI
Legal teams also need to ask one uncomfortable question. Why did the person spend an hour asking AI instead of coming to Legal? Sometimes the answer will be enthusiasm for the technology, but sometimes it may be because the business perceives Legal as difficult to access, slow to respond or requiring too much information before providing help. This friction is important to investigate.
If Legal wants employees to escalate issues before forming their own legal conclusions, the escalation route needs to be easier than the alternative. Structured legal intake can play an important role here. Instead of requiring an employee to decide whether they need a lawyer, the business manager can ask a series of guided questions about the commercial issue in a secure and governed tool. Risk indicators can trigger escalation. The correct intake tool can capture relevant information at the outset. Routine matters can continue through self-service pathways, while higher-risk matters reach the legal team earlier.
This approach allows the organisation to maximise the benefits of AI while balancing legal and governance concerns.
Final Thought
The General Counsel’s role has not become less important.
Generative AI has made legal information easier to access and interpret, but that does not make legal judgement less valuable. Arguably, it makes the ability to distinguish information from judgement even more important. General Counsel bring something an AI-generated answer cannot reproduce simply by producing more text, an understanding of the organisation, its history, its obligations, its commercial priorities and the consequences of the decisions being made.
For GCs contending with this new dynamic, the objective should not be to defend a professional identity by shutting AI out, but to protect the integrity of decision-making. That means giving the business appropriate freedom to use AI while defining clear boundaries for when exploration becomes an issue, when the legal team must be involved, and what information can safely be used in the process.