Phala Re-poster
Empowering 10,000 builders and companies to build scalable, private, and safe intelligence.
0 Följer
5 Följare
2 Gilla-markeringar
0 Delade
Inlägg
Alla
Videor
Our team put OpenClaw to the test across a 5-turn coding conversation and 8 real tasks. The primary insight from this benchmarking is quite revealing regarding expenses. OpenClaw typically does not incur high costs due to generating excessive text. Rather, the expense arises because the model persistently re-reads the background context.
Phala has established a partnership with Intel focused on Trust Authority. We are upgrading the attestation abilities of our Confidential AI infrastructure by adopting Intel Trust Authority (ITA). This solution acts as an independent third-party appraiser operating entirely apart from the Phala infrastructure.
The release of OpenClaw created an immediate sensation. It introduced an AI assistant capable of retaining context, executing tasks, and handling end-to-end workflow automation. Despite the hype, early attempts at configuration revealed serious vulnerabilities, such as the potential for complete agent hijacking through a simple browser tab. These security and usability hurdles remained until the Clawdi team successfully deployed the solution on Phala's TEEs.
🦞
🦞
Phala is experiencing sustained growth on OpenRouter, notably achieving a routing volume of 3.28B tokens on Feb 10 alone. At present, the platform features a diverse lineup of Confidential AI models, every single one of which runs inside TEE-secured infra. The available options include gpt-oss-120b, Qwen2.5 7B Instruct, and GLM 4.7 Flash, along with Gemma 3 27B, Qwen3 VL 30B A3B Instruct, and GLM 5.
Phala and Zeko Labs Unite
We are delighted to team up with @ZekoLabs to integrate confidential rollups designed for AI agents into the Phala TEE infrastructure. By hosting the sovereign, private rollup stack from Zeko within our hardware-enforced TEEs, we ensure that execution, key management, and sequencer operations are protected at the hardware layer.
We are delighted to team up with @ZekoLabs to integrate confidential rollups designed for AI agents into the Phala TEE infrastructure. By hosting the sovereign, private rollup stack from Zeko within our hardware-enforced TEEs, we ensure that execution, key management, and sequencer operations are protected at the hardware layer.
During an appearance on the Partnerships for Profit podcast, Jayson McQuown highlighted a critical issue regarding the interaction between machine learning and personal privacy. Phala's Director of Sales noted that while artificial intelligence expands its capabilities by training on your information, the question remains regarding who holds authority over that data. He emphasized that some information should remain protected, remarking that there are certain things that we probably don't want in the wild, wild web.
The highly anticipated GLM-5 by @Zai_org is finally here. A 744B-parameter open-source model built for complex systems engineering and long-horizon agentic tasks.
Phala is proud to partner with Zhipu AI as an official deployment partner, helping bring GLM-5 to production.
Phala is proud to partner with Zhipu AI as an official deployment partner, helping bring GLM-5 to production.
Enhancing Security at dstack: Hardening the Attestation Pipeline
Our team has officially released a suite of improvements designed to render attestation verification secure-by-default. We have fully resolved and deployed solutions for all identified findings. Furthermore, we have confirmed that there is no evidence of these vulnerabilities having been exploited. This update has been applied automatically, meaning that dstack users, downstream integrators, and partners do not need to take any action.
Our team has officially released a suite of improvements designed to render attestation verification secure-by-default. We have fully resolved and deployed solutions for all identified findings. Furthermore, we have confirmed that there is no evidence of these vulnerabilities having been exploited. This update has been applied automatically, meaning that dstack users, downstream integrators, and partners do not need to take any action.
As highlighted by The Quantum Insider @QuantumDaily, the fusion of security risks regarding AI and quantum technologies is occurring much more rapidly than predicted. This places enterprises in a position where they face immediate vulnerabilities from AI systems. Such a landscape proves why hardware-level security is essential. TEEs effectively isolate AI workloads at the processor level, which facilitates cryptographic attestation today and lays the groundwork for quantum-resistant security in the future.
Jayson McQuown, Phala Director of Sales, delivered a crucial message regarding artificial intelligence at #NTLSummit Miami. He cautioned that operating AI without a formidable security plan could potentially void attorney-client privilege. In his session, McQuown explored the significance of TEEs and confidential AI, making it clear that for healthcare, law, and similar privacy-critical industries, knowing the precise location of your data is an absolute necessity.
The dstack Community Call is making a comeback this Thursday. Participants will receive development and dstack updates directly from Phala CTO Hang. We are also hosting lightning talks by Tplus and Uniswap Labs. The event will include live demonstrations of @redpill_gpt as well as a privacy-preserving Telegram bot. Finally, join the Phala and Flashbots teams for open office hours.
Discover how to launch an AI agent featuring an ERC-8004 identity inside a TEE in less than 5 minutes by following our newly released tutorial.
Phala Partners with Storacha
By joining forces with @storachanetwork, Phala is working to advance the capabilities of AI agents. This collaboration leverages Storacha as the decentralized, encrypted storage layer, complemented by Phala providing TEE-backed confidential computation. It is a mutually beneficial result. 🍵 🤝 🐔
By joining forces with @storachanetwork, Phala is working to advance the capabilities of AI agents. This collaboration leverages Storacha as the decentralized, encrypted storage layer, complemented by Phala providing TEE-backed confidential computation. It is a mutually beneficial result. 🍵 🤝 🐔
The era where companies could simply say "trust us" has come to a close. In the current climate, the market demands proof that can be verified. Highlighting this shift at #Davos2026, the ETH Zurich President stated, “Trustworthy AI has evolved from an abstract aspiration to an operational necessity.”
https://www.weforum.org/stories/2026/01/scaling-trustworthy-ai-into-global-practice/
https://www.weforum.org/stories/2026/01/scaling-trustworthy-ai-into-global-practice/
We are making our way to #NEARCON2026!
Phala CEO @marvin_tong is set to be in San Francisco on Feb 23 & 24. He is prepared to meet with researchers and builders who support our vision: incorporating AI into the real economy while guaranteeing it stays trustworthy, open, and accessible to everyone.
http://nearcon.org
Phala CEO @marvin_tong is set to be in San Francisco on Feb 23 & 24. He is prepared to meet with researchers and builders who support our vision: incorporating AI into the real economy while guaranteeing it stays trustworthy, open, and accessible to everyone.
http://nearcon.org
.@nova_sdk is developing privacy-first, decentralized file sharing for user-owned AI.
By integrating Phala's verifiable TEEs via Shade Agents, encryption keys remain fully off-chain, hardware-secured, and attestable, unlocking safe AI data persistence at scale.
By integrating Phala's verifiable TEEs via Shade Agents, encryption keys remain fully off-chain, hardware-secured, and attestable, unlocking safe AI data persistence at scale.
According to the Check Point CTO at the World Economic Forum, AI is projected to contribute nearly $20 trillion to global GDP by around 2030—but only if it is built on verifiable trust. Consequently, the concept of trust defining AI’s $20 trillion economic opportunity is being featured as a major theme at this year’s @wef Annual Meeting in Davos.
With 19 U.S. states enforcing comprehensive privacy laws and AI-specific rules emerging, compliance is becoming a minefield, particularly around data-in-use protection.
Phala simplifies this by processing AI workloads inside verifiable TEEs, delivering data-in-use protection out-of-the-box.
Phala simplifies this by processing AI workloads inside verifiable TEEs, delivering data-in-use protection out-of-the-box.
With comprehensive privacy laws now enforced by 19 U.S. states and AI-specific rules emerging, compliance has become a minefield, particularly concerning data-in-use protection.
Phala Network simplifies this landscape by processing AI workloads inside verifiable TEEs, providing data-in-use protection out-of-the-box.
Phala Network simplifies this landscape by processing AI workloads inside verifiable TEEs, providing data-in-use protection out-of-the-box.
We're excited to partner with @DataHaven_xyz to power end-to-end confidential AI.
Phala secures computation with TEEs and trusted enclaves, while DataHaven provides private, verifiable storage for sensitive AI data and state. Together, we give builders a trusted path to deploy AI on highly sensitive data.
Phala secures computation with TEEs and trusted enclaves, while DataHaven provides private, verifiable storage for sensitive AI data and state. Together, we give builders a trusted path to deploy AI on highly sensitive data.
Logga in för att utforska mer innehåll