Goodix adds fingerprint plus secure element option for IoT and AI devices

Three tiers of fingerprint authentication

The portfolio covers fingerprint Match-on-Host (MoH), Match-on-Chip (MoC) and the new combined Fingerprint MoC + eSE solution. Goodix says it is built on capacitive, optical and ultrasonic fingerprint sensing, and is meant to help makers of IoT and AI devices offer faster, more flexible and more secure identity authentication.

The company frames the launch around a broader trend: as AI phones and PCs, wearables, robots, smart-home products and intelligent cockpits become more capable, secure identity authentication is becoming a foundational capability for edge devices.

With MoC, the whole biometric pipeline runs inside the fingerprint module: image capture, feature extraction, template storage and matching. Goodix says this complies with the Secure Device Connection Protocol (SDCP) and supports unique per-device keys that help protect both fingerprint data and its communication with the host processor. Because matching happens on the module, work is offloaded from the host processor, which the company says suits lightweight, long-battery-life devices that need fast response.

MoC + eSE: keeping credentials away from the host

For applications that need a higher level of security assurance, Goodix pairs a high-performance eSE chip with its fingerprint MoC solution. Assets such as API keys and account login tokens are separated from the host processor and stored in a trusted eSE chip, which the company says is certified to internationally recognized security standards.

The fingerprint MoC sends match results to the eSE over an encrypted channel, so the eSE can verify that the interaction was authorized by the legitimate user. The security solution also supports FIDO2 authentication and a full implementation of TLS 1.3. API keys and OAuth tokens are managed inside the secure element throughout their lifecycle, and communications are encrypted end to end.

Goodix describes the result as a zero-trust architecture on the device: "never trust, always verify." When credentials are used to reach cloud-based LLMs, they travel over an encrypted channel from the secure element to the cloud and are never exposed in plaintext to the host processor, according to the company.

The eSE also supports security policy enforcement. This includes biometric user authorization for high-risk operations and tamper-resistant audit logging, intended to prevent AI agents from altering the records.

Sensor options for different device designs

Goodix says its fingerprint technologies are designed to fit the varied form factors of AI and IoT devices:

• Capacitive: supports flat, curved, round, square and narrow-strip sensors, giving designers more flexibility.
• Optical: works beneath OLED displays and with a variety of transparent cover materials, is compatible with IR-transmissive ink and supports larger sensing areas.
• Ultrasonic: aimed at screenless devices, with FDT (finger-press detection) technology. Users press a designated area on the cover for fast recognition and unlocking, even with wet or oily fingers, according to Goodix.

What it means for device makers

The announcement targets manufacturers building AI agents and connected devices that handle sensitive credentials locally. The tiered approach lets them choose between host-based matching, on-chip matching, or on-chip matching combined with a secure element, depending on the security level and resources a product requires.

Goodix cites IDC, which says China's market for secure AI agent applications will reach US$1.6 billion by 2028, growing at a CAGR of more than 230%. The company says it will keep deepening the integration of biometric and security technologies with customers and partners for AI smartphones and PCs, AI wearables, smart homes, robots and intelligent cockpits. The release gives no pricing or availability details.