AI is changing what paralegals do rather than whether firms need them. Tools now handle first-pass document review and research, while paralegals handle verification, judgment, and client contact. Learning to use these tools well, and to check what they produce, has become a core paralegal skill rather than a bonus one.
If you’re considering a paralegal career, you’ve probably heard conflicting messages about artificial intelligence. Some headlines predict robots will take over legal work. Others dismiss AI as overhyped. The reality is more nuanced and, for prepared paralegals, more promising than either extreme suggests.
AI is already changing how legal work gets done. Document review that once took weeks now happens in days. Legal research that required hours of manual searching can be completed in minutes. These efficiency gains aren’t eliminating paralegal jobs. They’re shifting what paralegals do and which skills matter most.
This guide cuts through the noise with current data on AI adoption, an honest assessment of job market impacts, and practical steps you can take to build a career that thrives alongside AI rather than being threatened by it.
How Law Firms Are Using AI Right Now
AI adoption in law firms accelerated faster than anyone predicted. The ABA’s 2024 Legal Technology Survey found that 30.2% of respondents said their offices were using AI-based technology, up from 11% the year before. Clio’s Legal Trends Report put the figure higher still, with 79% of legal professionals reporting AI use, up from 19% in 2023. Don’t read those two numbers as one trend line. The surveys used different questions, samples, and methods, so they measure related but separate things.
Adoption tracks firm size. The ABA survey put usage at 47.8% among firms with 500 or more lawyers, while solo practitioners trailed at about 18%. Smaller practices are often where the gap is widest between what a firm has bought and what its staff can actually do with it.
The most common AI platforms in legal work include Thomson Reuters CoCounsel, which now serves over 20,000 law firms and most Am Law 100 firms. LexisNexis offers Lexis+ AI for research and analysis. Clio Duo brings AI assistance to practice management at accessible pricing. Harvey AI has captured significant market share among large firms willing to pay premium rates.
What are these tools actually doing? Among legal professionals who reported using generative AI in Thomson Reuters research, 77% used it for document review, 74% for legal research, and 74% for document summarization. Those are shares of the people already using AI, not of the whole profession. Thomson Reuters has put the expected time saving at roughly five hours a week.
One caution before you start experimenting at work. Plenty of firms have no written AI policy, or have one their staff have never seen. Ask before you put client material into any tool. A firm without a written policy still carries confidentiality obligations, and you don’t want to be the test case.
What Paralegals Actually Use AI For
Adoption statistics tell you that firms are buying these tools. They don’t tell you what the work looks like. Here’s where AI shows up in a paralegal’s actual day.
- First drafts of routine correspondence. Client letters, scheduling emails, confirmation letters. You edit rather than start from a blank page.
- Document templates. Complaints, motions, interrogatories, and discovery requests built from a structure you then adapt to the case.
- Summarizing long records. Deposition transcripts, medical records, and produced document sets condensed to something you can scan before reading the parts that matter.
- Case chronologies. Pulling dates and events out of a document set and ordering them, which is one of the most time-consuming manual tasks in litigation support.
- First-pass research. Finding a starting point in unfamiliar territory. This is a lead, not authority, and every citation has to be pulled and read.
- Clause and term spotting. Scanning a contract set for specific provisions, indemnity language, or termination terms.
- Plain-language rewriting. Turning dense legal text into something a client can follow.
Notice the pattern. AI does the volume work and produces a draft. A paralegal decides whether the draft is right. That division is where the job is heading, and it’s why verification skills matter more than tool familiarity.
What AI Can and Can’t Do in Legal Work
Understanding AI’s limitations matters as much as knowing its capabilities. AI excels at pattern recognition across large document sets. It can scan thousands of contracts and identify specific clauses in minutes. It can search case law databases and surface relevant precedents faster than any human. It can draft initial versions of routine documents based on templates and parameters.
These capabilities are genuine and valuable. But AI has critical weaknesses that make human paralegals essential.
The most serious problem is “hallucination,” where AI generates plausible-sounding but completely fabricated information. In legal work, this means fake case citations, invented statutes, and fictional legal principles presented with complete confidence.
The scale of this is easy to underestimate. Legal researcher Damien Charlotin maintains a public database of court and tribunal decisions in which the court found, or clearly implied, that a party relied on hallucinated material. In May 2025 it held roughly 120 cases. On August 31, 2026 it listed 1,996, of which 1,364 came from United States courts. It updates daily, so treat that as a floor and check the live count.
Its inclusion bar is narrow. A case gets in only when a court addressed the AI use in a written decision, which means the database records caught and documented incidents rather than every piece of inaccurate AI output filed anywhere.
It isn’t only self-represented litigants getting caught. On that same date the database recorded 794 entries where the party using AI was a lawyer, alongside 1,149 for self-represented litigants. The lawyer column is the one worth noticing. Those are regulated professionals who filed AI output without checking it.
The consequences are real and they’re escalating. In the landmark Mata v. Avianca case in 2023, attorneys submitted a brief citing six completely fabricated cases generated by ChatGPT. When the court questioned the citations, one attorney asked ChatGPT to confirm they were real. It did. The result was $5,000 in sanctions and a requirement to notify every judge falsely cited. Outcomes recorded in the database since then run from warnings and admonishments through five-figure monetary sanctions, adverse costs orders, bar referrals, struck briefs, and removal from a case.
Hallucination isn’t the only limit, and the real ones are narrower than the usual framing suggests. AI can discuss an ethical question competently. What it can’t do is carry professional responsibility for the answer. It can produce sympathetic text, but the trust that gets a frightened client through a custody dispute is built with a person. Newer systems can react to new information and chain several steps together, though someone still has to decide whether the chain went anywhere sensible. Witness credibility is a call for the people in the room. In each case the binding constraint is accountability rather than raw capability, and accountability is what a trained human supplies.
The ABA’s Formal Opinion 512, issued July 29, 2024, was the ABA’s first formal ethics opinion on generative AI. It speaks to lawyers under the Model Rules, covering competence, confidentiality, client communication, supervision, and fees, so it doesn’t set rules for paralegals directly. What it does is make a supervising lawyer answerable for AI-assisted work product, and Model Rule 1.1 expects lawyers to understand the benefits and risks of the technology they use. Somebody has to do that checking. On most legal teams, that somebody is a paralegal.
The Real Impact on Paralegal Jobs
The Bureau of Labor Statistics projects little or no change in paralegal employment from 2025 to 2035. Employment is expected to move from 404,900 in 2025 to 403,800 in 2035, a change of 0% and a net loss of 1,100 positions over the decade.
The BLS is unusually direct about why. Law firms are expected to remain the largest employer of paralegals, but demand is expected to be limited by advances in technology including artificial intelligence, which the agency expects to make paralegals more efficient at conducting research and preparing documents. For a deeper look at these trends, see our paralegal job outlook guide.
This headline number sounds alarming, but doesn’t tell the whole story.
Despite that flat projection, about 40,800 paralegal openings are expected each year on average through 2035. The BLS attributes most of those openings to the need to replace workers who transfer to other occupations or leave the labor force, including retirements. That’s worth stating plainly. These are replacement openings rather than newly created positions, which describes a different labor market than a growing one, but they are still real jobs that firms need to fill every year.
What paralegals earn
The national median paralegal salary was $62,890 in May 2025, which works out to $30.24 per hour. The mean sits higher at $69,700 per year, or $33.51 per hour. A mean above the median tells you the pay range is wide, with higher earners pulling the average up.
| Pay percentile | Hourly wage | Annual wage |
|---|---|---|
| 10th percentile | $21.51 | $44,740 |
| 25th percentile | $24.20 | $50,340 |
| Median (50th percentile) | $30.24 | $62,890 |
| 75th percentile | $38.50 | $80,080 |
| 90th percentile | $48.80 | $101,500 |
Read those as pay distribution, not as a career ladder. A long-tenured paralegal in a low-paying market can sit near the 10th percentile, and a relatively new one in a major metro can sit at the 75th. Geography and employer type move these figures as much as experience does.
Both columns matter. Paralegal compensation is almost evenly divided between hourly and salaried pay. NALA survey data found that 51% of paralegals were salaried, while about 49% were paid hourly. Being paid hourly doesn’t mean freelancing. Most hourly paralegals are firm employees on an hourly rate.
What about a pay premium for AI skills? Job postings increasingly ask for it, and recruiters and vendors publish salary-premium figures for AI-capable legal staff. Reliable paralegal-specific evidence for a consistent premium is thin, and the published ranges come from job-posting samples rather than from wage surveys. Read them as industry signals. The BLS figures above remain the benchmark for what the work actually pays.
The bottom line: AI isn’t eliminating paralegal careers. It’s raising the bar for which paralegals thrive and which struggle.
Skills That Matter in the AI Era
The paralegals who succeed in an AI-enhanced workplace combine technical proficiency with distinctly human capabilities.
On the technical side, prompt engineering has become a gateway skill. Getting valid results from AI tools requires knowing how to structure queries, provide context, and refine outputs. The difference between a vague prompt and a well-crafted one can mean the difference between useless output and genuine time savings.
How to write a prompt that actually works
Most disappointing AI output traces back to a thin prompt. Five habits fix most of it.
- Give it the source material. Paste the document. Don’t ask the tool to recall a case or a statute from memory, because recall is where fabrication happens most. Grounding a model in the actual text lowers that risk without removing it, so the output still gets checked.
- Name the role and the reader. A summary for a supervising attorney and a summary for a client need different length, tone, and detail.
- State the output format. Ask for a numbered chronology, a two-paragraph summary, or a table with named columns. Vague instructions produce vague shapes.
- Set the constraints. Jurisdiction, date range, word count, reading level. Constraints are what turn a generic draft into a usable one.
- Ask what it couldn’t determine. Requesting an explicit list of gaps and assumptions surfaces the weak spots instead of burying them in confident prose.
Compare “summarize this deposition” with “summarize this deposition transcript for a supervising attorney in under 300 words, list every date mentioned in a separate chronology, and flag any question the witness declined to answer.” The second one is a usable work product. The first one is a coin flip.
Verification and quality control skills matter enormously. Every AI-generated citation needs to be checked against authoritative databases. Every AI summary needs to be compared with the source documents. Every AI draft needs review for accuracy, tone, and jurisdiction-specific requirements. Paralegals who develop systematic verification processes become essential safeguards against the hallucination risks that have led to so many sanctions.
Proficiency with specific platforms adds value. Familiarity with legal research tools like Westlaw and Lexis (including their AI features), e-discovery platforms like Relativity, contract analysis software, and practice management systems all strengthen your marketability. Most of that stack predates the AI wave, and knowing the software already running in most firms gives you the foundation these AI features are being bolted onto.
But technical skills alone aren’t enough. The human capabilities AI cannot replicate are becoming more valuable, not less. Critical thinking allows you to spot when AI output doesn’t make sense in context. Emotional intelligence helps you navigate sensitive client situations. Communication skills let you explain complex legal concepts clearly. Judgment calls on strategy, ethics, and case direction remain firmly in human territory.
Practice area shapes how the work changes, though no labor-market source ranks paralegal specialties by how automatable they are. Treat what follows as a pattern rather than a forecast. Document-heavy work such as e-discovery, contract review, and corporate due diligence sees the heaviest AI use, because high volumes of standardized material are what these tools handle best. Family law, criminal defense, personal injury, and immigration carry plenty of automatable research and drafting too. What differs is the share of the job that sits in client contact, factual judgment, and advocacy, which runs higher in those areas and is harder to hand off.
AI Training and Certification Options
One necessary clarification: despite some marketing claims, neither NFPA nor NALA currently offers a dedicated “AI-Certified Paralegal” designation. Both associations provide AI-focused continuing education and conference sessions, but their primary certifications don’t yet include AI-specific competency tracks.
That said, legitimate training options exist and are worth pursuing.
| Program | Provider | Cost | Duration |
|---|---|---|---|
| Legal AI Fundamentals Certification | Clio | Free | About 2.5 hours, five modules |
| AI Boot Camp for Legal Professionals | Paralegal Boot Camp | $497, or six payments of $93 | About 6 hours of video, one year of access |
| GenAI for Paralegals: Streamlining Legal Drafting | Coursera | Free to enroll, certificate costs extra | About 3 hours |
Clio’s free certification offers the fastest path to foundational AI literacy. The five video modules cover what AI is, how to write effective prompts, cybersecurity considerations, and tool selection. You can finish it in an afternoon and add a shareable certificate to your resume. Clio also runs Emerging and Advanced tiers for people already using AI regularly.
Paralegal Boot Camp runs four modules across about six hours of video, plus monthly group coaching calls and a year of access. Coursera’s is the shortest of the three at roughly three hours, and it sits inside a broader AI-in-law specialization if you want to keep going.
One thing to watch. University continuing-education offerings in this space have been thinning out, so the options above are vendor and platform courses rather than academic programs. Duke Continuing Studies ran a 40-hour course called Embracing AI for Legal Professionals that still appears in plenty of roundups, and Duke has since discontinued its professional certificate programs. Confirm that any program you find is still enrolling before you build a plan around it.
If you hold a certification, don’t assume an AI course counts toward continuing education. NALA requires Certified Paralegals to complete 50 CLE hours per five-year recertification period, including at least five hours of legal ethics, and it caps technology credits at three hours. Clio says outright that its Fundamentals certification is not CLE eligible. Check your certifying body’s current rules and confirm a course is approved before you count on it. The training is still worth doing on its own terms, and it positions you for AI-specific credentials as professional standards evolve.
How to Future-Proof Your Paralegal Career
Start with free resources. Clio’s certification costs nothing and takes a few hours. Microsoft, Google, and AWS all offer educational content on AI fundamentals. YouTube has extensive tutorials on specific legal AI tools. Begin building familiarity before investing in paid programs.
Develop verification expertise. Create personal checklists to review AI output. Practice comparing AI-generated summaries with source documents. Build habits of skepticism that protect you and your firm from hallucination risks. This verification capability is your competitive advantage.
Strengthen human skills deliberately. Take communication courses. Practice explaining complex concepts in plain language. Develop your ability to manage client anxiety and build trust. These capabilities become more valuable as routine tasks shift to AI.
Document your impact. Track metrics showing how AI tools save time, reduce errors, or improve outcomes. Quantify your value contribution. These data points matter for performance reviews, raise conversations, and job searches.
Stay current on ethics and compliance. Follow court rulings on AI disclosure requirements. Understand your jurisdiction’s emerging standards. The ABA’s Formal Opinion 512 is essential reading. Position yourself as someone who understands not just how to use AI, but how to use it responsibly.
Frequently Asked Questions
How is AI changing the paralegal job?
AI is shifting the work rather than removing it. Routine tasks like initial document review and first-pass research are increasingly automated, while work requiring judgment, client interaction, and strategic thinking remains human territory. The BLS projects about 40,800 annual paralegal openings through 2035 despite flat overall employment. Paralegals who adapt to working alongside AI will find opportunities. Those who resist learning these tools may struggle to remain competitive.
What AI tools should paralegals learn first?
Start with general-purpose AI assistants like ChatGPT or Claude to learn how large language models behave and to practice prompt writing. Use non-confidential or practice material for that, and don’t put client or firm-sensitive information into any tool your employer hasn’t approved for that purpose. Then explore the legal-specific tools your firm uses or plans to adopt, such as CoCounsel, Lexis+ AI, or Clio Duo. Familiarity with e-discovery platforms like Relativity is valuable for litigation-focused roles. Focus on understanding AI capabilities and limitations rather than mastering any single tool, since the landscape changes rapidly.
Do I need an AI certification to get hired as a paralegal?
AI certification is not currently required for paralegal positions, and no widely recognized AI-specific paralegal credential exists yet. Traditional certifications, such as NALA’s Certified Paralegal, remain more important in hiring decisions. That said, demonstrating AI familiarity on your resume increasingly differentiates you from other candidates. Completing Clio’s free Legal AI Fundamentals Certification or similar training shows initiative and positions you as a forward-thinking candidate.
Which paralegal specializations are most affected by AI?
Document-intensive practice areas see the heaviest AI use. E-discovery, contract review, corporate due diligence, and intellectual property research involve high volumes of standardized documents, which is what these tools handle best. No labor-market source ranks specialties by how automatable they are, so treat this as a pattern rather than a forecast. Family law, criminal defense, personal injury, and immigration also contain automatable research and drafting, but a larger share of the job sits in client contact, factual judgment, and advocacy. Consider your interests and strengths when choosing a specialization.
How can I convince my firm to let me use AI tools?
Start by understanding your firm’s concerns, which typically center on confidentiality, accuracy, and ethics compliance. Propose a pilot project with clear boundaries: a specific task, a defined scope, and verification protocols. Emphasize that you’ll review all AI output before it goes anywhere. Reference the ABA’s Formal Opinion 512 to show you understand ethical requirements. Document time savings and quality outcomes from your pilot. Success with a small project builds trust for expanded AI use.
Key Takeaways
- AI adoption in law firms climbed sharply in a single year, with the ABA’s 2024 survey putting office use at 30.2% and Clio’s research putting broader legal-professional use at 79%. The two surveys measure different things and shouldn’t be read as one trend.
- The BLS projects about 40,800 annual paralegal openings through 2035 despite 0% employment change, driven by workers leaving the occupation rather than by new positions, and names AI as a reason demand stays limited.
- A public database of court decisions involving AI-fabricated material grew from roughly 120 cases in May 2025 to 1,996 by August 31, 2026, with 794 involving lawyers rather than self-represented litigants, which is why verification skills matter.
- Paralegal pay ranges from $44,740 at the 10th percentile to $101,500 at the 90th, with a median of $62,890 per year or $30.24 per hour, and compensation splits almost evenly between hourly and salaried.
- No major “AI-Certified Paralegal” designation exists yet, and not every AI course carries CLE credit, so confirm approval with your certifying body before counting on it.
- Human skills, including judgment, empathy, client relationships, and strategic thinking, become more valuable as routine tasks shift to AI.
Ready to Start Your Paralegal Career?
Explore accredited paralegal programs that prepare you for success in an AI-enhanced legal workplace. Find the right education path for your goals.
May 2025 US Bureau of Labor Statistics salary figures for Paralegals and Legal Assistants. National 2025-35 employment projections from the US Bureau of Labor Statistics Employment Projections program. Figures reflect national data, not school-specific information. Conditions in your area may vary. Data accessed August 2026.
