Ad Fraud Statistics 2026: The Numbers Every UK Marketer Should Know

If you buy digital media, a meaningful share of your budget is reaching machines, not people. These are the numbers worth committing to memory in 2026 — each with its source and year, and what it means for your spend.

How much of the internet is even human?

• 51% of all web traffic is now automated (Imperva, 2025 Bad Bot Report) — for the first time in a decade, non-human activity outweighs human.
• Bad bots make up 37% of internet traffic, up from 32% the year before (Imperva, 2025) — the malicious share is climbing, driven by generative AI.
• 45% of all bot attacks are now simple, high-volume attacks, up from 40% in 2023 (Imperva, 2025) — AI has lowered the barrier to entry, so crude attacks are being mass-produced alongside sophisticated ones.
• 44% of advanced bot traffic is directed at APIs (Imperva, 2025) — evidence that attackers increasingly go after business logic rather than the front door.
• Travel is the most bot-attacked sector (Imperva, 2025) — a warning for any advertiser in a high-value, high-intent category.

What ad fraud costs

• Global ad-fraud losses are projected at $100.2bn in 2026 (Juniper Research, 2023–2028 forecast) — up from around $63bn in 2025.
• Losses are forecast to hit $172bn by 2028 (Juniper Research, 2023–2028 forecast) — the problem compounds year on year.
• Around 20.6% of programmatic ad traffic is invalid (Juniper Research, 2023–2028 forecast) — roughly one in five impressions.
• The fraud-detection industry itself will top $4.8bn in 2026 (Juniper Research, 2023–2028 forecast) — the market spends billions just to police the problem.

Where the waste hides in programmatic

The ANA’s 2023 Programmatic Media Supply Chain Transparency Study remains the most-cited audit of where open-web programmatic money actually goes. It traced $123m of spend across 35.5 billion impressions.

• Only ~36% of programmatic spend reaches impressions that are non-MFA, brand-safe, measurable and viewable (ANA, 2023) — the rest is absorbed by unqualified inventory and the “cost waterfall”.
• As much as $20bn of the $88bn open-web programmatic market is waste (ANA, 2023) — around 23% of the total.
• Made-for-advertising sites took 15% of programmatic spend and 21% of impressions (ANA, 2023) — an estimated $13bn a year flowing to pages built only to carry ads.

Is anyone actually fixing it?

Partly — and the most recent data suggests the gains are fragile.

• Participating marketers cut MFA spend from 15% to 6.2% (ANA Programmatic Benchmark Study, 2024) — with the median falling from 10% to 1.1%, proof the problem responds to attention.
• Advertisers still wasted $26.8bn on programmatic in a single quarter (ANA Q2 2025 Programmatic Transparency Benchmark) — progress on MFA has not closed the wider gap.
• MFA exposure rose to 1.1% in Q1 2026, after holding between 0.4% and 0.6% through 2025 (ANA Q1 2026 Programmatic Transparency Benchmark) — the first increase since 2023.
• The best-performing advertisers convert 54.0% of programmatic spend into qualified impressions; the worst manage 32.1% (ANA, Q1 2026) — a 21.9-point spread, the widest the benchmark has recorded. Who you buy with now matters as much as what you buy.

Connected TV is the soft target

• CTV invalid traffic hit 18% globally in Q1 2025, up from 12% a year earlier — and 24% in the UK (Pixalate, 2025) — UK advertisers face a materially worse rate than the global average.
• Bots account for 65% of all fraud in CTV (DoubleVerify, 2025 Global Insights) — 14 points higher than other channels.
• Around 4 million infected devices generate fake CTV traffic every day (DoubleVerify, 2025 Global Insights) — an industrial-scale, always-on supply of fraudulent impressions.
• CTV fraud schemes surged 140% globally (DoubleVerify, 2026) — growth the company attributes directly to AI-assisted scheme creation.

The AI accelerant

Almost every worsening number in this list traces back to the same cause. Generative AI has made fraudulent inventory and the traffic to fill it cheap to produce at scale.

• AI-generated “slop” content accounts for 1.3–2.4% of open-web programmatic ad spend (TAG/ANA/Fiducia, 2026) — the first attempt to size it, and already comparable to the industry’s entire MFA exposure.

That last figure is the one to watch. MFA took a decade to become a $13bn problem; AI slop reached a comparable share of spend within roughly two years of generative tools becoming widely available.

What the numbers add up to

The pattern is consistent: reported reach and real, human reach are drifting further apart, and the supply chain absorbs the difference. Three implications follow for anyone planning 2026 budget.

Averages hide the spread. The ANA’s 21.9-point gap between the best and worst advertisers means the industry-wide figures above are not your figures. Some advertisers are losing far less than the average, and some far more, and nothing in a platform dashboard tells you which you are.

Post-campaign reporting is too late. Every number here describes money already spent. Filtering before the impression is bought is what changes the outcome; a monthly report only measures the damage.

The only number that matters is your own. Verifying delivery at the device level is the only way to know which half of your budget is working.

Treat this page as a living scorecard, not a one-off read. The individual figures shift each quarter, but the direction of travel is fixed: as AI makes bots cheaper to build and the supply chain harder to see into, the distance between what you’re billed for and what a human actually saw keeps widening. The marketers who stay ahead aren’t the ones who memorise every stat — they’re the ones who measure the gap in their own campaigns and act on it.

If you’d like that number for your own media, we’ll measure it on a live campaign.