AI Due Diligence Statistics (2026)

84%

84% of senior deal professionals use AI in early-stage or outside-in diligence, and 78% use it in confirmatory diligence, including the Q&A process.

56%

56% of organisations deploy agentic AI in due diligence and valuation, and 53% in deal sourcing and strategy.

50%

50% of dealmakers have AI regularly embedded in due diligence, and 62% say human-only decision-making is no longer defensible in complex transactions.

70%

70% of US private equity buyout professionals rank diligence synthesis among the top three areas where AI raises productivity. Respondents report an average 28% productivity improvement from AI use.

40%

40% of private equity GPs say generative AI's return on investment is highest in due diligence, across legal, financial and commercial work, ahead of 33% for deal sourcing and market mapping.

46%

46% of companies invest in generative AI for pre-deal M&A work in 2026, up from 31% in 2024. For post-deal work the share is 27%, up from 18%.

45%

45% of M&A executives now rely on AI tools, after adoption more than doubled in 2025.

95%

95% of private equity funds report AI initiatives meeting or exceeding their original business case.

24%

24% of private equity leaders report revenue growth from AI, and 5% have fully integrated AI into operations, against 14% across all industries.

86%

86% of UK financial firms use AI, up from 75% in 2024 and 58% in 2022, and every responding UK and international bank uses it. Generative AI accounts for 56% of the AI systems firms report.

80%

80% of UK financial firms have AI agents deployed and a further 14% plan to within a year, but only 2% give agents broad autonomy. Most deployed agents recommend actions for people to decide on and carry out.

85%

85% of banks supervised directly by the ECB use generative AI, and more than 90% have integrated AI into their operations. Risk mitigation, led by fraud and cybercrime prevention, is the most important use, followed by customer support and credit scoring.

45%

45% of EU securities market firms have no AI use case in production or in development, rising to 53% of investment firms and investment managers. Of 342 firms reporting on the challenges of deploying AI, 51% cite hallucinations.

65%

Almost 65% of European insurers already use generative AI and a further 23% plan to within three years. Data privacy and security, regulatory compliance and a lack of skilled staff are the most significant reported challenges.

50%

Around 50% of Swiss financial institutions use AI or have initial applications in development, and a further 25% intend to within three years. 91% of those using AI also use generative AI.

6%

Only 6% of asset managers say they use AI for decision-making, although 55% have integrated it into at least one strategy's investment process. 69% name data quality or access as the main barrier to further adoption.

95%

95% of alternative investment fund managers use generative AI in their work, up from 86% in 2023.

75%

Active use of AI in finance functions rose to 75% from 30% in 2024, and 36% of organisations name data quality as both their biggest barrier and their biggest opportunity.

33.9%

33.9% of US finance and insurance businesses used AI in a business function in the past two weeks, against 19.8% of all US businesses, as of 3 May 2026.

40%

40% of professionals say their organisation already uses generative AI, up from 22% a year earlier. The share is 41% at law firms and 47% in corporate legal teams.

74%

Among legal professionals using generative AI, 74% use it for document review, 80% for legal research, 73% for summarising documents and 49% for contract drafting.

92%

92% of legal professionals use at least one AI tool, and 62% report weekly time savings of 6% to 20%. The mean weekly time saving across respondents is nearly 10%.

240hours

Legal professionals expect AI to free up nearly 240 hours a year each, about five hours a week, worth an average of $19,000 per professional.

33%

Leading AI legal research tools hallucinated on between 17% and 33% of queries: Westlaw AI-Assisted Research on 33%, Lexis+ AI and Ask Practical Law AI on 17%.

75.1%

On data extraction from legal documents, Harvey Assistant scored 75.1% and CoCounsel 73.2%, above the lawyer baseline of 71.1%. Vincent AI scored 69.2% and Oliver 64.0%.

80×

The AI tools in the same study were between 6 and 80 times faster than the lawyers, but the lawyers beat every tool on redlining, scoring 79.7% against 53.6% to 65.0%.

78%

AI legal research products scored between 74% and 78% on 200 US legal research questions, against a lawyer baseline of 69%.

2,289

Courts had issued 2,289 decisions involving AI-hallucinated content as of 10 November 2026, 910 of them involving lawyers and 1,552 in the United States.

69.47%

The best AI model scored 69.47% on Vals AI's Finance Agent benchmark of entry-level analyst research tasks over SEC filings. On the benchmark's financial modelling tasks, the best score was 57.0%.

73.19%

The best AI model answered 73.19% of Vals AI's expert questions on credit agreements correctly. The agreements in the test often run past 200 pages.

77.38%

The best AI agent scored 77.38% on expert Excel modelling tasks such as LBO, DCF and data room summaries, each at least five hours of work for an expert. Vals AI reads high scores as substantial but incomplete work rather than client-ready deliverables.

48.9%

Retrieval problems caused 48.9% of the failures when 20 AI models answered questions from company annual reports, the largest single cause. The study attributes 44.7% of failures to comprehension errors and 9.1% to standalone hallucinations.

60%

AI models failed approximately 60% of realistic analyst tasks modelled on work at hedge funds, private equity firms and investment banks, and scored under 5% on questions that required their own assumptions.

40%

40% of senior deal professionals reported hallucinated AI outputs leading to inaccurate diligence in the past 12 months.

81%

GPT-4 Turbo with a shared retrieval system answered incorrectly or refused to answer 81% of questions about public-company filings, mostly through refusals. With the whole filing in its context window it answered 79% correctly.

25.1%

Consultants using GPT-4 completed 12.2% more tasks and finished them 25.1% faster, with significantly higher quality, on tasks within the model's capability. On a task outside it, they were 19% less likely to reach a correct solution.

15%

Customer support agents with a generative AI assistant resolved 15% more issues per hour on average, and less experienced agents 30% more. The most skilled agents saw small gains in speed and small declines in quality.

40%

Professionals using ChatGPT took 40% less time on writing tasks, and the quality of their output rose 18%.

47.6%

Expert graders rated 47.6% of Claude Opus 4.1's deliverables as better than or as good as work by industry professionals, across 220 tasks in 44 occupations, including financial and investment analysts.

1.39×

Used naively, GPT-5 was about 90 times faster than the experts in the same study. Once the time to review and redo its work was counted, the gain fell to between 1.12 and 1.39 times faster.

26%

Equity research reports linked to FactSet's generative AI drew on 26% more distinct information sources, covered 24% more topics and used 21% more analytical methods. Relative forecast accuracy fell when analysts faced heavier information-processing demands.

58%

58% of sell-side analysts use AI at least a little, 55% cite concerns about accuracy or bias, and fewer than 3% use it mainly to build independent forecasts.

130%

Law students using AI tools were between 50% and 130% more productive on five of six legal tasks, measured as quality points per minute. A contract task was the exception.

3%

Workers who adopted AI chatbots reported saving about 3% of their work hours, and the study rules out effects on earnings or hours larger than 2%.

75%

Financial planning and analysis teams spent 75% of their time gathering data and administering processes, down from 77% in 2010.

4

The Financial Stability Board names four AI-related vulnerabilities that could raise systemic risk: third-party dependencies and provider concentration, market correlations, cyber risks, and model risk, data quality and governance.

$400k

In its first AI washing cases, in March 2024, the SEC settled charges against two investment advisers, Delphia and Global Predictions, for false claims about their use of AI, with $400,000 in total civil penalties. Later actions include a settled case against Presto Automation in January 2025.

2027

Under the EU's AI Omnibus, in force since 27 July 2026, the AI Act's rules for the high-risk systems listed in Annex III apply from 2 December 2027, and those for AI in regulated products from 2 August 2028. The rest of the Act has applied since 2 August 2026.

78%

78% of AI users bring their own AI tools to work, and the share is 80% at small and medium-sized companies.

48%

48% of employees report uploading company information, such as financial, sales or customer data, into public AI tools, and only two in five say a policy guides their use of generative AI.

82%

82% of investment industry employers say a lack of standards hinders faster adoption of AI, and data privacy and security is the most commonly named biggest roadblock, cited by 16%.

54%

The top three model and API providers accounted for 54% of the model relationships UK financial firms reported, and the top three cloud and compute providers for 79% of cloud relationships. Third-party implementations made up 39% of reported AI systems and 53% of high-materiality ones, up from 33% and 35% in 2024.

40%

Concerns about information security or operational resilience were the most common major constraint on adopting third-party AI, cited by 40% of UK financial firms, ahead of limited transparency into third-party systems at 34%. 42% see UK data protection and privacy rules as a high compliance burden.

24%

24% of UK financial firms rely on deep knowledge of the third-party AI systems they rate as high materiality, against 77% for systems built in-house. For third-party systems, 81% rely on vendor assurance and contracts instead.

94%

94% of UK financial firms have firmwide AI policies, principles or standards, and 84% have central AI or model risk governance forums. Ongoing monitoring, leakage testing and periodic review of AI data risks were each reported by about half.

80%

80% of financial services leaders name data privacy and security as one of the AI risks that concern them most, ahead of legal, intellectual property and regulatory compliance at 55%. 85% of their organisations are increasing AI investment.

6,622

A data room for a $100 million to $500 million deal held about 6,622 files and 109 users. For a $50 million to $100 million deal it was about 3,325 files, and for a $1 million to $50 million deal about 1,206.

237hours

Deal teams spent an average of 237 hours in the data room per deal in the first half of 2026, up from 219 hours a year earlier.

170days

The global median diligence time was 170 days in the third quarter of 2026, down from 174 days a year earlier: 235 days in Asia Pacific, 171 in the Americas and 155 in EMEA.

253days

Deals closed in the first half of 2026 took an average of 253 days from start to close, about 4% shorter than a year earlier and the shortest since 2022.

49%

49% of US private equity dealmakers reported longer deal timelines than six months earlier, against 16% who reported faster execution.

42%

42% of US mid-market business leaders named the length or complexity of due diligence as a barrier to completing deals, and 52% named the quality or completeness of the target's financials.

46%

46% of M&A professionals named due diligence issues as the most prevalent contract or process factor in 2025 M&A disputes, up from 37% a year earlier.

81%

81% of the professionals in the same survey said their involvement in M&A disputes rose in 2025.

29%

About 29% of private-target deals saw the buyer's initial purchase price adjustment calculation disputed after closing.

18%

About 18% of representations and warranties insurance policies bound from 2019 to 2023 have had at least one claim notification. Breaches of financial statement warranties account for 38% of paid losses.

33%

33% of private equity GPs said diligence red flags, such as earnings quality or customer churn that did not hold up, were the top reason deals they pursued did not close in 2025. A valuation gap of more than 10% was cited by 36% and being outbid by 25%.

3.1%

3.1% of US public-target deals announced in 2025 were terminated, against 1.6% in 2024. The median target termination fee was 2.7% of transaction value, and 56.3% of deals carried a reverse termination fee.

24%

Earnouts appeared in 24% of private-target deals in 2025, excluding life sciences, against a historical norm closer to 15%. Among US private equity buyers the share was also 24%, against 13% from 2014 to 2022.

21%

Sellers received about one in five potential earnout dollars, roughly 21% of the maximum, and 59% of earnout deals paid something.

10.0%

The median total escrow in 2025 was 10.0% of transaction value: 11.1% in deals without representations and warranties insurance and 2.8% in deals with it.

63%

63% of private-target acquisition agreements referenced representations and warranties insurance, up from 55% in the previous study.

$91.6bn

Marsh placed a record $91.6 billion of transactional risk insurance limits globally in 2025, up 34%, across more than 3,800 policies. North American primary representations and warranties rates rose 16%.

$904bn

Global buyout deal value rose 44% to $904 billion in 2025, while the number of deals fell 6% to 3,018.

$717bn

Global buyout-backed exit value jumped 47% to $717 billion in 2025, while the number of exits fell 2% to 1,570.

32,000

About 32,000 unsold companies worth $3.8 trillion sit in private equity portfolios.

7years

Holding periods at exit hover around seven years, and almost 40% of portfolio companies have been held for more than five years, up from 29% in 2019.

$1.3tn

Global buyout dry powder stands at $1.3 trillion, and it is getting older.

$395bn

Buyout fundraising fell 16% to $395 billion in 2025, while total private capital raised was flat at about $1.3 trillion.

14%

Distributions from buyout funds stayed at 14% of net asset value, below 15% for the fourth year running.

$1.5tn

Global private equity investment reached $1.5 trillion in the first nine months of 2026, on pace for the third-best year on record, while deal volume was on pace for the slowest year since 2020.

$461bn

US private equity deal value totalled $461 billion in the first half of 2026, down 10.6% on the first half of 2025.

12.5×

The median US buyout entry multiple closed 2025 at 12.5 times EV/EBITDA, up from 9 times in 2015 and level with the 2021 peak.

$118bn

Secondary market volume reached a first-half record of $118 billion in 2026, up 15%, and GP-led deals made up $62 billion of it.

$4.6tn

Global M&A value reached $4.6 trillion in 2025, up 49% on 2024 and the highest since 2021, with 68 mega deals, the most since records began in 1980.

$3.9tn

Global M&A reached $3.9 trillion in the first nine months of 2026, up 28%, while the number of deals fell 8% to its lowest since 2020.

$891bn

Sponsor-backed M&A reached a record $891 billion in the first nine months of 2026, up 35%.

49%

49% of deal projects on Datasite's platform reached completed status in the third quarter of 2026, up three percentage points: 54% in the Americas, 56% in Asia Pacific and 39% in EMEA.

2,006

2,006 transactions were reported under the Hart-Scott-Rodino Act in fiscal 2025, against 2,031 in fiscal 2024, and 31.8% were valued at more than $1 billion.

2.1%

US antitrust agencies issued second requests on 2.1% of reportable transactions in fiscal 2025, down from 3.0% in fiscal 2024. For deals above $10 billion the rate was 21.1%.

12.3months

Significant US merger investigations took an average of 12.3 months from announcement to conclusion in 2025, an all-time high. In the first three quarters of 2026 the average was 10.1 months across 13 investigations.

$4.1tn

The global private credit market reached $4.1 trillion in assets under management at the end of 2025, growing 17% for the second year in a row.

$190bn

Private credit funds held final closes on $190 billion in the first half of 2026, 53% more than in the first half of 2025. Full-year 2025 closes totalled $240 billion.

$1.5tn

Private credit lenders had extended about $1.5 trillion of loans to US businesses at the end of 2025, and business development companies held about $570 billion in assets.

$36.9bn

Direct lenders refinanced $36.9 billion of broadly syndicated loans in 2025, while $34.1 billion of direct-lender loans moved the other way into the syndicated market. Borrowers that moved to the syndicated market cut their spread by 226 basis points on average.

5.0×

The median private credit borrower carried leverage of 5.0 times, against 3.2 times for leveraged loan borrowers, with median revenue of $223 million against $902 million and average interest cover of 2.2 times against 3.7 times.

20%

Perpetual-life BDCs and interval funds held about 20% of net assets in private credit vehicles, $161 billion and $80 billion. In the first quarter of 2026, accepted redemptions from these semi-liquid vehicles exceeded inflows for the first time since they were created.

50%

Semi-liquid vehicles accounted for more than 50% of new private credit fundraising in the US in 2025, before the recent wave of redemptions.

$1.2tn

US insurers held about $1.2 trillion of private credit at the end of 2025, about 13% of their cash and invested assets and 21% of their bond holdings.

$800bn

Private equity funds and business credit intermediaries, which include private credit funds and BDCs, held $1.3 trillion of bank credit commitments and were drawing more than $800 billion in early 2026.

$23bn

NAV loans rated by KBRA reached a record $23 billion across 38 deals in 2025.

2.41%

Proskauer's private credit default index stood at 2.41% in the third quarter of 2026, down from 2.51% in the second quarter.

10%

Payment-in-kind interest featured in 10% of private market loans in the third quarter of 2026. In 6% of all loans it was "bad PIK", added after a deal that closed without PIK.

8%

Payment-in-kind interest has risen to roughly 8% of public BDCs' gross investment income, from about 5.5% in 2021.

10.6%

10.6% of loans in private credit managers' portfolios had a significant amendment, such as a payment holiday, a maturity change, a covenant change or a PIK amendment, down from 11.65% in 2024.

21%

21% of the direct lending deals Proskauer closed in 2025 were covenant-lite, and 91% of those were for borrowers with EBITDA above $50 million.

2.75%

Median non-accrual investments at KBRA-rated BDCs rose to 2.75% of investments at cost in the second quarter of 2026, from 1.81% in the first quarter.

2.5%

The size-weighted covenant default rate on private market loans was 2.5% in the third quarter of 2026, down from 2.7% in the second quarter and below the six-year average of 3.8%.

45cents

Implied recoveries on restructured direct loans fell to 45 cents on the dollar in 2025, from 51 cents in 2024. Loans that had paid some interest in kind recovered 41.6 cents, against 52.6 cents for the rest.

$287bn

About $287 billion of US leveraged loans mature in 2028, and about half of it was issued by borrowers rated B-minus or lower.

146bps

The spread gap between syndicated and direct lending buyout loans narrowed to 146 basis points in 2026 so far, the narrowest since 2019.