AI Safety Startup Funding 2025-2026

What are all the funding deals in the AI safety market from August 2025 to July 2026?

The table below lists every disclosed equity round raised by pure-play AI safety companies between August 2025 and July 2026. We count as pure-play AI safety companies those focused on reducing AI harms and failure modes through evaluation, red teaming, guardrails, safety monitoring, risk platforms, robustness, governance, security, privacy controls, or explainability when those capabilities are designed for AI-specific risk reduction.

Each row shows the company, what it does, its category, the deal date, the funding stage, the round size, the region, the main investors when specified in the dataset, and the announcement source. For a wider view of how this market is developing, we cover it in our AI Safety market report.

How active has fundraising been in the AI safety market?

As of July 2026, fundraising in the AI safety market has been active but uneven. Over the past 12 months, pure-play AI safety companies raised 27 disclosed equity rounds and $660.62M combined.

Deal flow averaged 2.45 rounds per month, but the median month had only 2 deals. That means the AI safety market is visible, but not yet broad enough to produce steady monthly volume.

Capital flow was even more uneven. Average monthly funding was $60.06M, while median monthly funding was $21.17M, showing that a few large months created most of the dollar signal.

June 2026 was the clearest acceleration point. It produced 8 deals and $207M raised, which made it the busiest month by deal count and one of the two largest months by capital.

If you want to go deeper on company formation and funding pace, see our AI Safety market report.

How concentrated has fundraising been in the AI safety market?

As of July 2026, fundraising in the AI safety market is highly concentrated at the top. Over the past 12 months, the top 1 deal accounts for 22.71% of disclosed capital, the top 3 reach 44.50%, and the top 5 reach 62.37%.

The largest deal was LMArena’s $150M Series A. That single financing was bigger than the entire Guardrail Platforms category and bigger than every disclosed deal outside North America combined.

The top 10 deals account for 82.35% of all disclosed capital. This means the AI safety market’s dollar total is mostly a story about a small set of platform-scale companies.

This concentration matters because it changes how totals should be read. A rising market total does not necessarily mean that every AI safety subcategory is getting easier to fund.

How much of the AI safety funding signal is driven by outliers?

As of July 2026, a large share of the AI safety funding signal is driven by outliers. Over the past 12 months, 5 deals of $50M or more accounted for a major share of the market’s total capital.

Megarounds above $50M represent only 18.52% of disclosed deals. Yet the top 5 deals alone account for 62.37% of disclosed capital, which shows how strongly large rounds shape the market narrative.

The average round size was $24.47M, while the median was only $10M. That gap confirms that the average round should not be read as the typical AI safety financing.

Excluding rounds above $50M, the AI safety market raised $248.62M. That smaller number is often the better baseline for understanding ordinary company formation.

Is the AI safety market broad with many targets, or narrow with few fundable companies?

As of July 2026, the AI safety market is broadening, but it is still narrow by venture-market standards. Over the past 12 months, the dataset includes 27 deals across 26 unique companies.

Runlayer is the only company that appears twice, with both a Seed round and a Series A during the period. That shows repeat financing exists, but most activity still comes from first visible rounds.

The market is not evenly spread across all safety functions. AI Risk Platforms account for 12 deals, while Model Robustness Tools show no disclosed qualifying rounds in the dataset.

The buyer problem is broad, but the fundable company set is still forming. Investors are backing governance, agent control, evaluation, and runtime enforcement faster than deeper technical robustness.

Is AI safety mostly an early-stage formation market or a late-stage scaling market?

As of July 2026, the AI safety market is mostly an early-stage formation market by deal count, but not by capital. Over the past 12 months, Seed rounds accounted for 16 of 27 disclosed deals, or 59.26%.

Seed capital totaled $160.42M, which equals 24.28% of disclosed funding. That is meaningful formation activity, but it is not where most dollars landed.

Series A rounds captured $356M, or 53.89% of all disclosed capital. This shows that once an AI safety company reaches commercial traction, round sizes can reprice quickly.

Late-stage activity was sparse but large. Only 2 Series B rounds appeared, yet they represented $138M and 20.89% of total capital.

For more context on how stage labels map to market maturity, read our deeper analysis of the AI safety market.

Which categories attract the most investor attention in AI safety?

As of July 2026, AI Risk Platforms attract the most investor attention in the AI safety market by deal count. Over the past 12 months, they produced 12 of 27 disclosed deals, or 44.44%.

AI Risk Platforms raised $230.5M, or 34.89% of disclosed capital. This category includes governance, agent access, visibility, secure deployment, compliance firewalls, and enterprise control layers.

AI Evaluation Tools lead by capital, with $232.95M raised from only 3 deals. LMArena and Braintrust show that evaluation can raise platform-scale rounds when it becomes critical market infrastructure.

AI Guardrail Platforms also show broad investor attention, with 7 deals and $95.37M raised. The category is active, but its capital share remains below its deal share.

Which categories attract disproportionately large checks in the AI safety market?

As of July 2026, AI Evaluation Tools attract the most disproportionately large checks in the AI safety market. Over the past 12 months, the category captured 35.26% of capital from only 11.11% of deals.

The capital-share to deal-share ratio for AI Evaluation Tools is 3.17x. That means evaluation has the clearest large-round signal, even though it is not the most crowded category.

AI Safety Monitoring is close to proportional, with a ratio of 0.97x. Its $94.8M across 4 deals suggests solid but less extreme investor conviction.

AI Risk Platforms have a ratio of 0.79x, while Guardrail Platforms sit at 0.56x. Both categories are active, but their average checks are smaller than their deal counts suggest.

If you want to understand where capital is concentrating by product layer, explore our market report covering AI safety categories.

Which geographies matter most for fundraising in the AI safety market?

As of July 2026, North America matters most for fundraising in the AI safety market. Over the past 12 months, the region captured $546.3M, or 82.70% of disclosed capital.

North America also led by deal count, with 14 of 27 disclosed rounds. But its capital share is much larger than its deal share, which shows a clear valuation and round-size premium.

Europe ranked second with 5 deals and $73.95M raised. Its 18.52% deal share versus 11.19% capital share suggests credible formation, but smaller financing depth.

Asia-Pacific also produced 5 deals, but only $20.37M in capital. The region has technical formation, yet its median deal size of $4M is far below North America’s $24M median.

Is the AI safety opportunity set broad or concentrated in one hub?

As of July 2026, the AI safety opportunity set is geographically concentrated, but not in only one hub. Over the past 12 months, North America dominated dollars, while Europe and Asia-Pacific both showed visible company formation.

North America holds 51.85% of deals and 82.70% of capital. That makes it the funding base for the AI safety market, not just the largest region.

Europe and Asia-Pacific each produced 5 deals, equal to 18.52% of deal count for each region. The difference is that Europe raised $73.95M, while Asia-Pacific raised $20.37M.

The Middle East appeared through 3 deals and $20M raised. Latin America and Africa were absent from the disclosed AI safety dataset, which shows how concentrated the public startup market remains.

Is AI safety a market of small experiments or scaled financings?

As of July 2026, the AI safety market is still mostly a market of small and mid-sized experiments, with a few scaled financings on top. Over the past 12 months, 18 of 27 deals were below $20M.

The largest bucket is $5M to less than $20M, with 12 deals. Another 6 deals were below $5M, which confirms that much of the market remains early and exploratory.

At the same time, 5 deals were $50M or more. Those larger rounds show that some AI safety companies have already moved from tool-building into platform-scale positioning.

The median round size of $10M is the best shorthand for the typical deal. The average of $24.47M is useful, but it is pulled upward by LMArena, Braintrust, Straiker, Arcade.dev, and WitnessAI.

We cover this split between experiments and platform candidates in more detail in our full market deck on AI safety.

Who are the investors that appear the most in AI safety fundraising?

As of July 2026, repeat investor activity in the AI safety market is visible but still limited. Over the past 12 months, only a small group of investors or affiliated groups appeared more than once.

Andreessen Horowitz or a16z-affiliated vehicles appear in 3 deals if grouped across LMArena, Braintrust, and ZeroDrift. Counting only strict Andreessen Horowitz entity appearances lowers that number to 2.

Felicis appears in 2 deals, LMArena and Runlayer Series A. Khosla Ventures or Keith Rabois also appear in 2 deals, across Runlayer’s Seed and Series A rounds.

Mirae Asset group appears across AIM Intelligence and Tynapse, while Samsung-affiliated entities appear across WitnessAI and AIM Intelligence if grouped at the corporate-affiliate level. These repeats cluster around evaluation and agent-control infrastructure.

One important caveat is that investor announcements rarely disclose individual check sizes. Investor repetition should therefore be read as participation frequency, not as exact dollars committed.

Who is the author of this content?

NEW MARKET PITCH TEAM

We track new markets so founders and investors can move faster

We build living "market pitch" documents for emerging markets: AI, synthetic biology, new proteins, and more. Instead of outdated PDFs or hallucinated LLM answers, our clients get a clean, visual, always-updated view of what's really happening: key players, deals, regulations, and signals that matter. Learn more about us.