Three independent security research teams successfully compromised fully patched Google Pixel 10 smartphones during Pwn2Own Ireland 2026, demonstrating that even devices running the latest security updates can contain exploitable vulnerabilities. The attacks took place on October 8 in Cork, Ireland, where researchers competed to identify weaknesses in smartphones, connected devices and enterprise technologies. One of the successful demonstrations earned Ikotas Labs a $300,000 prize, the largest individual award available for the smartphone category, while the three Pixel 10 compromises collectively generated $562,500 in rewards.
The results offer an important perspective on the security of modern mobile devices. Smartphones have evolved into highly sophisticated computing platforms protected by hardware-backed security mechanisms, application sandboxing, memory protections and increasingly restrictive operating systems. Yet their complexity also creates opportunities for vulnerabilities to emerge across browsers, wireless communication components and system services. The Pixel 10 demonstrations illustrate that security updates significantly reduce known risks without necessarily eliminating vulnerabilities that researchers have yet to discover or manufacturers have not fully addressed.
Three Teams, Three Successful Remote Attacks
The first successful demonstration came from researchers Tim Becker and Yves Bieri of Xint, who compromised the Pixel 10 using a vulnerability already known to the competition organizer or vendor. Under Pwn2Own rules, this situation is classified as a collision, meaning the vulnerability overlaps with an existing discovery. Although the attack was successful, the collision reduced the team’s reward to $150,000 and 15 competition points, half the maximum available for the category. The result nevertheless demonstrated that the vulnerability remained relevant to a fully updated device at the time of the competition.
Ikotas Labs followed with a more complex attack involving multiple vulnerabilities chained together to achieve the required compromise. The team received the full $300,000 award and 30 points, despite its entry also being classified as a collision in the published results. The organizers did not explain why that classification did not reduce the prize, nor did they disclose the technical details of the exploit chain. The achievement contributed to Ikotas Labs winning the competition’s overall Master of Pwn title, recognizing the team that accumulated the highest number of points across eligible targets.
The third successful attack was demonstrated by Dimitrios Valsamaras, Ken Gannon and Tenia Valsamara, who combined two vulnerabilities to compromise the smartphone. Their exploit chain included one previously known vulnerability and one zero-day, earning the researchers $112,500 and 22.5 points. Together, the three demonstrations accounted for three successful compromises out of four registered remote attempts against the Pixel 10. The remaining attempt, conducted during the competition’s opening day, was unsuccessful because the researchers ran out of time rather than because the organizers confirmed that the attack was technically impossible.
All three successful entries belonged to the remote exploitation category, which permits attacks through malicious web content opened in the device’s default browser or through supported wireless interfaces, including Wi-Fi, Bluetooth, NFC and cellular baseband communications. Researchers were required to demonstrate meaningful compromise, such as executing attacker-controlled code or extracting sensitive information from the target. However, the exact entry points, exploit mechanisms and capabilities demonstrated by each team have not been publicly disclosed. Consequently, the results should not be interpreted as confirmation that attackers can compromise arbitrary Pixel 10 devices remotely without user interaction.
What the Results Reveal About Smartphone Security
The distinction between remote exploitation and a completely interaction-free attack is particularly important. A vulnerability triggered through malicious web content may require a user to visit a specially crafted page, while a weakness in a wireless component could potentially be exploited through an entirely different mechanism. Both scenarios fall within the competition’s remote attack category, but they involve different prerequisites, exposure levels and defensive considerations. Without technical documentation describing the successful exploit chains, it is impossible to determine which scenarios apply to the demonstrated Pixel 10 compromises.
Another significant detail is the role of vulnerability collisions. At least two of the three successful Pixel 10 attacks incorporated previously known weaknesses, while the third explicitly combined an existing issue with a newly discovered zero-day. This illustrates how attackers and security researchers can achieve meaningful results by combining vulnerabilities rather than relying on a single critical flaw. A weakness that appears limited in isolation may become considerably more dangerous when paired with another vulnerability that enables privilege escalation, bypasses security restrictions or provides access to sensitive components.
Modern mobile operating systems are specifically designed to make this kind of exploitation difficult. Their security architectures typically separate applications from privileged system processes, restrict access to sensitive information and introduce multiple defensive layers intended to prevent one compromised component from exposing the entire device. As a result, sophisticated attacks often require exploit chains that cross several security boundaries. The Pixel 10 results reinforce the importance of evaluating vulnerabilities within the context of the broader system rather than treating each software defect as an independent security problem.
The demonstrations also highlight a fundamental limitation of vulnerability management. Installing every available security update is essential, but a fully patched device is not necessarily free from exploitable weaknesses. Security updates address vulnerabilities that have been identified, investigated and corrected, while previously undiscovered flaws may remain present in otherwise current software. Google’s October Pixel security bulletin was published on October 6, two days before the successful Pwn2Own demonstrations, and did not include fixes associated with the competition. At the time of reporting, the organizers had not announced patches or device-specific mitigation instructions for the newly demonstrated attacks.
Samsung, AI Infrastructure and Connected Devices Also Fall
The Pixel 10 was not the only major target compromised during Pwn2Own Ireland. Samsung’s Galaxy S26 experienced an even more extensive series of successful demonstrations, with researchers compromising the smartphone in all seven registered attempts. Six of those seven attacks involved at least one vulnerability collision, and the organizers confirmed that one weakness used in an earlier Ikotas Labs exploit chain was already known to the manufacturer but remained unpatched at the time. These findings illustrate how competition results can expose not only previously undiscovered vulnerabilities but also weaknesses that vendors have identified without yet delivering complete remediation.
The competition extended well beyond smartphones, reflecting the growing diversity of devices and platforms that require advanced security research. Participants successfully targeted printers from Lexmark, Canon and Brother, alongside connected products such as the Sonos Era 300 smart speaker, Philips Hue Bridge Pro and Home Assistant Green. Researchers also demonstrated attacks against the Garmin Index BPM, a connected health device, while AI-related targets included OpenAI’s Codex coding agent and Oracle Autonomous AI Database. Ikotas Labs achieved successful demonstrations against both AI platforms, further contributing to its overall competition victory.
Across the three-day event, 51 of the 63 scheduled attempts succeeded, and the published awards exceeded $1.2 million. Every product included in the competition schedule was successfully exploited at least once, underscoring the breadth of the vulnerabilities uncovered. Although Apple’s iPhone 17 and WhatsApp were eligible targets with potential prizes reaching $300,000, no researchers registered attempts against either product. Their absence from the results therefore provides no meaningful basis for comparing their security with that of the Pixel 10 or Galaxy S26.
The inclusion of AI infrastructure is particularly noteworthy because it reflects how the cybersecurity attack surface is expanding beyond conventional operating systems and network appliances. AI coding assistants, inference infrastructure and database platforms increasingly operate with access to valuable information and automated workflows. Vulnerabilities in these environments can create risks involving sensitive data, application integrity and potentially privileged operations. Their appearance alongside smartphones and embedded devices demonstrates that the underlying challenge is no longer confined to a particular category of technology: software complexity and interconnected systems are creating security risks across almost every modern computing environment.
Why Responsible Disclosure Matters After the Competition
Pwn2Own is organized by Trend Micro’s Zero Day Initiative, which provides financial incentives for researchers to demonstrate vulnerabilities under controlled conditions and disclose their findings responsibly. Successful participants submit technical documentation and exploit information to the organizers, who coordinate with affected vendors to support remediation. Manufacturers generally receive a 90-day period to develop security updates before the initiative publicly releases detailed vulnerability information. This process is designed to encourage legitimate security research while reducing the likelihood that newly discovered weaknesses become immediately available to malicious actors.
The financial rewards also reflect the significant effort required to discover and exploit vulnerabilities in modern systems. Developing a reliable attack against a fully patched smartphone can involve extensive reverse engineering, vulnerability research, exploit development and testing against multiple defensive mechanisms. A successful demonstration lasting only a few minutes may represent weeks or months of preparation, particularly when the attack requires chaining several independent weaknesses. By compensating researchers for their findings, competitions such as Pwn2Own provide an alternative to markets where exploit capabilities may be acquired for surveillance, espionage or criminal purposes.
For Google and other affected manufacturers, the immediate priority is understanding the demonstrated weaknesses and developing appropriate fixes. Until technical details become available, there is limited value in speculating about specific exploitation techniques or assuming that the competition results represent active attacks against ordinary users. Maintaining current security updates, avoiding unnecessary application installations and exercising caution with untrusted content remain sensible precautions, but they should not be presented as guaranteed defenses against vulnerabilities whose technical characteristics have not yet been disclosed.
The broader lesson from Pwn2Own Ireland 2026 is that modern cybersecurity depends on continuous testing rather than confidence in any single protective mechanism. Smartphones now incorporate increasingly sophisticated defenses, yet researchers continue to identify ways around them by exploiting overlooked interactions between software components and security boundaries. The same pattern is visible across connected devices, enterprise applications and emerging AI platforms, where new functionality often introduces additional complexity that must be evaluated.
The three successful Pixel 10 attacks do not establish that Google’s flagship smartphone is fundamentally insecure, nor do they demonstrate that widespread exploitation is occurring. They show something more useful: even mature security architectures benefit from independent research capable of identifying weaknesses before they become part of real-world attacks. As mobile devices continue to serve as gateways to personal identities, financial information and enterprise resources, the ability to discover and remediate these vulnerabilities before adversaries exploit them remains one of the most important measures of effective security engineering.