Skip to content

    Draft – for author review

    Legal Operations

    Using AI for contract review: what it can and can't do

    Agred Tafaj

    Agred Tafaj, LL.M. (Cambridge)

    Published · Updated · 2 min read

    AI tools can now summarise a contract, flag unusual clauses and suggest redlines in seconds. For a busy founder, it's tempting to paste a contract into a chatbot and sign whatever it says is fine.

    Used well, AI is a genuinely useful part of contract work. Used carelessly, it creates new risks. Here's a practical view of where it helps and where human judgement still matters.

    What AI does well

    • Summarising long contracts into readable overviews
    • Finding specific clauses quickly, such as renewal, liability or governing law
    • Comparing a contract against a standard template or playbook
    • Extracting key data (dates, parties, values) for a contract register
    • Producing first-draft redlines on routine points

    These are tasks where speed and consistency matter, and where a human can easily check the output.

    Where AI falls short

    Commercial context. AI doesn't know your business, your risk appetite, your relationship with the counterparty or your negotiating leverage. Whether a clause is acceptable depends on all of those things.

    Accuracy. AI tools can misread clauses, miss how definitions change the meaning of an obligation, or confidently state things that aren't true. Interactions between clauses, such as an indemnity that sidesteps the liability cap, are easy to miss.

    Law and jurisdiction. General-purpose tools may apply concepts from other legal systems. English law has its own approach to issues such as consequential loss, penalties and implied terms.

    Accountability. An AI tool won't stand behind its review or explain its reasoning on a call with your counterparty.

    Confidentiality and data protection

    Before uploading a contract to any AI tool, check what happens to the data. Consumer chatbots may retain inputs or use them to improve their models. Contracts often contain confidential information and personal data, so sharing them may breach an NDA or your UK GDPR obligations.

    Prefer enterprise tools with clear terms on data retention, training and security, and check your contracts don't prohibit sharing the information with third-party processors.

    A responsible way to use AI

    • Use AI for first-pass triage and summaries, not final decisions
    • Give it a playbook of your preferred positions to compare against
    • Always have a person check outputs against the actual contract text
    • Keep a record of which tools are approved for which data
    • Train your team on the limits as well as the features

    Human judgement, supported by smart tools

    The most effective approach combines both: AI to handle volume and speed, and an experienced reviewer to apply commercial judgement and catch what the tool misses. That's the model I use and help clients adopt. [ADD SOURCE-BACKED STATISTIC]

    Need this checked in your own contract? Book a free 20-minute call.

    Book a call

    Agred Tafaj, LL.M. (Cambridge) is an independent legal consultant specialising in commercial contracts and legal operations, and the founder of ION-X. More about Agred

    This article is general information under the law of England and Wales and is not legal advice. ION-X is an independent legal consultancy, not a firm of solicitors. For help with your specific situation, get in touch.

    Related articles