Pwn2Own Ireland 2026 pays out $1.26 million for 98 zero-days
Pwn2Own Ireland 2026 wrapped on 9 October 2026 with 98 zero-day flaws demonstrated and $1,262,000 in prizes, led by Ikotas Labs. The leaderboard confirms the hunting ground is shifting toward mobile and AI infrastructure.
9 October 2026. Pwn2Own Ireland 2026 concludes with $1,262,000 paid out for 98 zero-day flaws. Ikotas Labs wins the edition with 42.5 Master of Pwn points and $361,000 over three days. 9 October 2026. The Google Pixel 10 falls three times on the final day, pushing the total to 98 flaws across 29 teams. Why it matters: the Zero Day Initiative contest is a forward indicator — the categories it chooses reveal where the industry expects risk, and this year that means mobile and AI infrastructure.
The leaderboard, or the concentration of talent
The numbers tell a story of concentration. Ikotas Labs dominates with 42.5 points and $361,000, chaining the Samsung Galaxy S26, OpenAI Codex and the Oracle Autonomous AI Database. The team also took the contest’s biggest single prize — $300,000 — by chaining multiple zero-days to compromise the Google Pixel 10 on the third day.
Xint took second place with $240,000 and 27.5 points. Team ZyGoat rounded out the podium with $125,000 and 27.5 points. The gap between first and second — $121,000 — reflects a recurring fact of the contest: a few teams capture most of the value, and the prize spread measures the depth of their exploit chains more than their number of targets.
The day-by-day run gives the sense of pace. On the first day, Interrupt Labs, Ikotas Labs and Nguyen Thanh Dat (Viettel Cyber Security) brought down the Galaxy S26, for $388,500 and 32 zero-days. On the second day, $232,500 rewarded 45 flaws, with PetoWorks, the KAIST Hacking Lab and a team combining Dimitrios Valsamaras, Ken Gannon and Tenia Valsamara (CENSUS Labs) taking down the S26 three more times. On the third day, $641,000 landed for 21 flaws.
Mobile and AI become the new hunting grounds
The 29 teams spread across seven categories: mobile phones (Galaxy S26 and Pixel 10), AI infrastructure, AI coding apps, messaging apps, smart home devices, printers, and a brand-new category for wellness healthcare devices.
Two categories dominate the story: mobile and AI. The Galaxy S26 and Pixel 10 were attacked repeatedly, a sign that flagship smartphones remain the richest target in prizes. But the presence of OpenAI Codex, the Oracle Autonomous AI Database and a dedicated “AI coding apps” category marks a turning point: AI infrastructure and agents are now treated as a standalone attack surface, on par with a browser or a kernel.
That is the takeaway for a CISO. Five years ago, an “AI” category did not exist in this contest. Today it produced the edition’s biggest prize. The vulnerability market — and therefore attacker effort — is shifting toward agents, vector databases and assisted coding pipelines.
The iPhone left on the table
The most paradoxical detail is what was not attacked. Apple put an iPhone 17 on the table, with a maximum award of $300,000 for a remote compromise. No team registered to attempt it.
Do not conclude that the iPhone 17 is invulnerable. The reason is economic: a remote iPhone compromise demands a long, expensive exploit chain, usually reserved for actors who would rather sell it quietly to a broker than demonstrate it publicly for $300,000. The emptiness around the iPhone says less about iOS security than about the market price of a remote exploit against that target.
That reading is useful for defence. The targets researchers choose to demonstrate in public are the ones with the most favourable effort-to-prize ratio — in other words, the ones where exploitation is ripest. The iPhone’s absence from the leaderboard should not reassure; it simply moves the worry toward the targets that actually fell.
What 98 zero-days mean for a defender
98 zero-day flaws in three days is a brutal reminder of the asymmetry: a single motivated team produces, over a weekend, more exploitable flaws than a vendor fixes in a release cycle. But the contest is not just a prize spectacle.
Every demonstrated zero-day is disclosed to the vendor, who has 90 days to ship a fix before ZDI publishes the details. The practical consequence is a deferred wave of patches: today’s flaws become public security bulletins in roughly three months. A defender who follows the leaderboard therefore knows in advance which products will receive critical fixes in the coming quarter.
Last year gives the trend. Pwn2Own Ireland 2025 demonstrated 73 flaws for $1,024,750; Summoning Team won with $187,500. The jump from 73 to 98 flaws and from $1.02 million to $1.26 million in a year measures the market’s acceleration, not just prize generosity.
The recommendation is simple: treat the leaderboard as an early warning. If your estate contains Galaxy S26s, Pixel 10s, AI coding agents or autonomous databases, expect imminent critical patches — and prepare your maintenance windows accordingly.
Verdict
If you run an Android mobile fleet, prioritise Samsung and Google patch tracking now: the Galaxy S26 and Pixel 10 were compromised repeatedly, and their fixes land within 90 days. If you deploy AI agents or AI infrastructure, add them explicitly to your attack-surface inventory and your patch cadence — the contest made them its most profitable category. If you follow the vulnerability market, read the leaderboard as a maturity thermometer: wherever the prizes are highest, attackers are already at work.