Cerebras IPO Surges 68% Amid OpenAI Data Breach Revelations
Cerebras grabs headlines with a successful IPO, OpenAI confirms a major data breach, and Osaurus rolls out new AI models for Mac users.
EDITION #153, May 15
Cerebras grabs headlines with a successful IPO, OpenAI confirms a major data breach, and Osaurus rolls out new AI models for Mac users.
Image source: CNBC
Big News from the Tech World: Cerebras Goes Public and OpenAI Faces Security Breach
Cerebras Systems made headlines with a blockbuster IPO on May 14, 2026, raising $5.55 billion and achieving a market cap near $100 billion. The company’s Wafer Scale Engine 3 chip, significantly larger than traditional GPUs, is designed for AI tasks and has attracted major clients like OpenAI, which signed a $20 billion deal in January 2026.
In contrast, OpenAI faced a security breach where hackers compromised two employees’ devices through a supply-chain attack on the TanStack library. While no user data was accessed, the incident highlights vulnerabilities in open-source software. OpenAI is taking precautionary measures, including rotating digital certificates.
- Cerebras' shares jumped 68% on debut, closing at $331.07 after an IPO priced at $185.
- TanStack's breach involved 84 malicious updates published in a six-minute window, detected within 20 minutes.
- Cerebras' WSE-3 chip is claimed to be 57 times larger than the largest GPU, enhancing AI processing.
- OpenAI is rotating digital certificates to mitigate risks following the recent security incident.
Image source: Bloomberg
IPO SURGE
Cerebras' IPO Soars 68%
Cerebras Systems made a spectacular market debut, closing its first day of trading up 68% from its IPO price of $185, reaching a valuation of nearly $95 billion. The AI chipmaker raised $5.55 billion, marking the largest U.S. tech IPO since Uber in 2019. This surge comes amid strong demand for its Wafer Scale Engine 3, which is designed to outperform traditional GPUs in AI inference tasks.
The IPO minted two new billionaires: CEO Andrew Feldman and CTO Sean Lie, who hold stakes valued at approximately $3.2 billion and $1.7 billion, respectively. Early investors, including Benchmark and Foundation Capital, also saw significant gains. Cerebras' impressive debut signals a robust appetite for AI hardware and could pave the way for future IPOs in the sector.
Source: TechCrunch
DATA BREACH ALERT
OpenAI Confirms Data Breach
OpenAI has confirmed that two employees’ devices were compromised in a recent security incident linked to the TanStack open-source library. While unauthorized access to internal source code repositories occurred, the company states that no user data or core systems were compromised. As a precaution, OpenAI is rotating digital certificates, which may require updates for macOS users.
This incident follows a troubling report from TanStack, which revealed that 84 malicious software versions were distributed in just six minutes, aiming to steal credentials. OpenAI's swift response highlights its commitment to security despite the breach.
AI ON MAC
Osaurus Introduces AI Models for Mac
Osaurus has launched an innovative open-source LLM server specifically for Mac users, allowing seamless switching between local and cloud AI models. With support for Apple’s on-device models and cloud providers like OpenAI and Anthropic, it addresses security concerns by keeping files and tools isolated on the user’s Mac. Since its launch, Osaurus has been downloaded over 112,000 times, indicating strong interest in AI solutions tailored for Mac environments.
The platform ships with over 20 native plugins, enhancing user experience and productivity. It also acts as a Model Context Protocol (MCP) server for compatible clients, making it a versatile tool for developers and tech enthusiasts alike.
Source: The Verge
AI DETECTION EXPANDED
YouTube Broadens AI Deepfake Detection Tool Access
YouTube is expanding its AI deepfake detection tool to all users aged 18 and older, enhancing safety on the platform. This feature, initially tested with select creators and officials, uses facial recognition to identify and alert users about potential lookalikes in videos. Users can request the removal of matched content, which is evaluated based on realism and identification criteria.
The rollout aims to combat misinformation and manipulated media, aligning with YouTube's commitment to user safety. Users can opt out and delete their data, ensuring privacy while tackling deepfake challenges.
DATA BREACH ALERT
Japan's Hotel Check-In System Exposed
A significant data breach involving Japan's hotel check-in system, Tabiq, has put over one million guests' personal information at risk. The incident, reported by security researcher Anurag Sen, stemmed from a misconfigured Amazon cloud storage bucket managed by Reqrea, which left passports, driver’s licenses, and selfie verification photos accessible to anyone.
As Reqrea investigates the breach, it highlights the urgent need for improved data security practices, especially in hospitality. Similar vulnerabilities in software supply chains, as seen in recent OpenAI incidents, further stress the importance of rigorous security audits and access controls to safeguard sensitive guest information.
