Tech Intelligence Copy RSS link Link copied to clipboard!
David Sacks says AI will reshape cyber offense and defense
David Sacks, a member of the US Council of Advisers on Science & Technology, stated that AI models are tools reshaping cybersecurity rather than doomsday devices. He noted that while AI discovers existing vulnerabilities, it can also help patch them, potentially hardening systems. Sacks emphasized the need for defenders to access these models before attackers, citing rapid development in China. He predicted a market shift towards a new equilibrium between AI-powered offense and defense.
Palo Alto Networks to acquire Portkey to boost AI security play
Palo Alto Networks announced its intent to acquire AI infrastructure startup Portkey to strengthen capabilities in securing enterprise adoption of autonomous AI systems. The deal is expected to close in the fourth quarter of fiscal 2026, subject to necessary approvals. Financial terms were not disclosed. Palo Alto Networks plans to integrate Portkey's AI gateway into its Prisma AIRS platform to address risks associated with AI agents. Portkey, founded in 2023, provides an LLMOps platform acting as a single access layer for over 1,600 AI models.
UK Cyber Strategy Lacks Regional Capability Infrastructure
Analysis of CYBERUK 2026 identifies a critical gap between national cyber strategy and execution, attributing it to missing regional capability infrastructure. Without this layer, SMEs struggle to scale, supply chains destabilise, and resilience fails to distribute effectively. The report argues that while government policy provides clear direction, the mechanisms to translate strategy into scalable organisational resilience are undefined, leading to uneven capability development and potential system fragility.
Samsung reports record quarterly profit driven by 49-fold jump in chip income
Samsung Electronics reported a record quarterly profit of 57.2 trillion won, with its chip division generating 53.7 trillion won, a 49-fold increase from the previous year. This surge is attributed to a boom in AI data centre construction and sustained spending by major technology firms. While Samsung secured multi-year contracts to manage supply constraints, it faces potential production disruptions from union strikes in South Korea. Additionally, rising component costs are expected to reduce profitability in the mobile and display divisions.
Data centre industry evaluates direct current architectures for AI workloads
The rapid expansion of artificial intelligence workloads is driving data centre operators to reconsider power architectures, with direct current (DC) emerging as a potential solution for high-density racks exceeding 200kW. While alternating current (AC) remains dominant, the industry is exploring high-voltage DC distribution, such as 800VDC, to improve efficiency and reduce cabling constraints. Experts note that no single standard has emerged yet, with a likely future involving hybrid AC/DC systems and coexistence of multiple configurations as technical standards are defined by initiatives like the Open Compute Project.
Apple CEO Tim Cook warns of extended memory crunch and higher costs
Apple CEO Tim Cook warned that rising memory costs will increasingly impact the business during a Q&A session following the company's earnings call. Citing supply constraints driven by AI infrastructure demand, Cook stated the company would evaluate a range of options to address the issue. While Apple reported revenue growth of 17% for the fiscal second quarter, executives noted that memory shortages will significantly affect Mac models in the upcoming quarter. The global memory crisis has forced tech giants like Microsoft and Meta to increase capital expenditure forecasts due to elevated component prices.
Multiple organisations suffer data breaches and cyber attacks in April 2026
In April 2026, numerous global entities including Mercor, Cisco, the European Commission, Jones Day, Snowflake, Eurail, Los Angeles City Attorney's Office, China's supercomputers, Basic-Fit, Booking.com, Rockstar Games, McGraw-Hill, Ameriprise, Bol.com, Vercel, Canada Life, Russia's Gonets system, France Titres, Rituals, Udemy, UK Biobank, Coupang, ADT, Medtronic, Vimeo, Pitney Bowes, Amtrak, and Movistar Peru experienced data breaches or cyber attacks. Threat actors such as TeamPCP, Lapsus$, Silent ransom group, ShinyHunters, FlamingChina, and World Leaks were involved. Incidents ranged from supply chain compromises and stolen source code to the exposure of personal, financial, and sensitive government data affecting millions of individuals across various sectors including healthcare, technology, finance, and public services.
Cybersecurity standards and supply chain risks emerge for EV charging infrastructure
EV charging networks face growing cyber threats due to grid connectivity, with vulnerabilities in software, hardware, and third-party apps. Recent data breaches, including one involving Tesla chargers in the Middle East, highlight risks to customer data and grid stability. Governments in the US and EU are tightening regulations, such as the NIS2 Directive and proposed bans on Chinese components, to mitigate supply chain risks. Industry bodies are promoting standards like OCPP 2.0.1, ISO 15118, and NIST IR 8473 to enhance security protocols, encryption, and network segmentation across the sector.
UK AI Security Institute finds GPT-5.5 matches Claude Mythos in cyber attack tests
The UK AI Security Institute (AISI) reports that OpenAI's GPT-5.5 performs on par with Anthropic's Claude Mythos Preview in cyberattack evaluations. GPT-5.5 fully completed a multi-stage enterprise attack simulation and achieved a 71.4% success rate on expert-level tasks, slightly edging out Claude Mythos. AISI notes these results indicate growing AI autonomy in security breaches. However, a universal jailbreak bypassed all of OpenAI's safety safeguards, and GPT-5.5 failed to solve an industrial control system simulation. The tests lacked active defenders, highlighting risks for poorly protected networks.
White House opposes Anthropic plan to expand access to Mythos AI model
The White House has opposed Anthropic's proposal to expand access to its Mythos AI model to approximately 70 additional companies. Officials cited national security concerns and the risk of misuse, noting potential vulnerabilities in critical infrastructure such as power plants and hospitals. The model is currently restricted to select partners including Amazon, Apple, Google, Microsoft, and NVIDIA. Concerns were heightened by reports of unauthorized access during the model's limited release phase.
Anthropic launches public beta of Claude Security for enterprise customers
Anthropic has launched the public beta of Claude Security, a vulnerability detection and remediation tool available to enterprise customers. Powered by the Claude Opus 4.7 model, the platform performs deep contextual code analysis to reduce false positives and provide actionable patch instructions. New features include scheduled scanning, directory-level targeting, and webhook integrations. Major cybersecurity vendors are embedding the underlying model into their tools.
BufferZoneCorp account used to distribute malicious Ruby gems and Go modules
A sophisticated supply chain attack originating from the GitHub account BufferZoneCorp has been uncovered, targeting developers and continuous integration environments. Threat actors deployed sleeper packages in the Ruby and Go ecosystems that impersonated legitimate tools. These malicious packages execute credential theft by harvesting environment variables and scanning for sensitive files like SSH keys and AWS credentials. The stolen data is exfiltrated to remote endpoints. Additionally, some Go modules manipulate GitHub Actions workflows by altering environment variables, disabling checksum verification, and planting fake execution wrappers. Certain payloads also establish persistent backdoor access by appending SSH public keys to authorized files. Security teams have blocked the identified Go modules, but Ruby gems remained active during early investigations. Experts advise auditing environments and rotating credentials.
Hyperscalers improve data center cooling efficiency with industrial instrumentation
Hyperscalers are enhancing data center cooling loop performance by adopting industrial-grade instrumentation, including 0xDN flowmeters and non-invasive temperature sensors. These technologies address spatial constraints and cybersecurity requirements while enabling tighter thermal control. Improved measurement accuracy reduces the need for over-cooling, lowering energy consumption and utility costs. Additionally, integrated diagnostics support predictive maintenance, preventing system failures and ensuring operational uptime amidst rising AI-driven power demands.
Intel stock doubles in April for best month in 55 years
Intel's stock surged 114% in April, marking its best month in 55 years on the Nasdaq. The rally follows a turnaround under new CEO Lip-Bu Tan, who replaced Pat Gelsinger in March 2025. Demand for Intel's CPUs has increased due to AI bottlenecks, with major customers like Google, Microsoft, and Amazon securing supply. The U.S. government holds a 10% stake in the company. Intel is also expanding its foundry business and advanced packaging capabilities, securing commitments from Tesla and SpaceX.
Convicted former Harvard scientist rebuilds brain computer lab in China
Charles Lieber, a former Harvard scientist convicted in 2021 of lying to U.S. authorities about Chinese ties, has established a new research lab in Shenzhen, China. Overseeing the state-funded i-BRAIN institute under the Shenzhen Medical Academy of Research and Translation, Lieber now has access to dedicated nanofabrication equipment and primate research facilities unavailable in the US. The lab focuses on brain-computer interfaces, a technology identified as a national priority by the Chinese government with potential military applications.
Qualcomm teases agentic CPUs and smartphones
Qualcomm CEO Cristiano Amon announced plans to supply custom hyperscale silicon to a leading hyperscaler, with shipments expected in the December quarter. The company is developing dedicated CPUs for agentic experiences in data centers and high-performance AI inference accelerators. Amon also previewed 'agentic smartphones' featuring integrated AI assistants, noting a shift towards more capable CPUs and increased memory requirements. Qualcomm expects to win 70 percent of Samsung's SoC business this year and next. For Q2, the company reported $10.6 billion in revenue and a 162 percent increase in net income to $7.37 billion. The CEO highlighted growth in the automotive sector, with over one million cars using Qualcomm processors for advanced driver assistance.
US Navy partners with Domino Data Lab for AI mine detection in Strait of Hormuz
The US Navy has partnered with AI firm Domino Data Lab to deploy artificial intelligence for detecting Iranian mines in the Strait of Hormuz. Supported by a nearly $99.7 million contract under Project AMMO, the initiative involves investing up to $100 million to train underwater drones. The technology aims to replace human-operated ships with AI-driven solutions to enhance speed and efficiency in identifying new mine types, thereby safeguarding global oil shipments and maritime safety in the region.
Keeper Security launches Agent Kit for AI coding agents
Keeper Security has launched Agent Kit, a tool designed to secure AI-driven developer workflows by preventing credential exposure in chat history or source control. The product integrates Keeper Secrets Manager and Keeper Commander with coding agents such as Claude Code, Cursor, Codex, and GitHub Copilot. It enables agents to retrieve secrets for local runtime use and administer vault resources while maintaining role-based access controls and audit logging. Available as open-source software under the Apache 2.0 licence, the kit addresses risks associated with embedding AI assistants into coding tasks. The launch targets organisations extending internal security rules to automated software agents.
Cyber experts take an optimistic view of AI-powered hacking
At the Centre for Emerging Technology and Security (CETaS) conference, Alexander Babuta of the Alan Turing Institute expressed optimism regarding Anthropic's Claude Mythos Preview, noting its potential for enterprise vulnerability detection and patching. Conversely, Adam Beaumont of the AI Security Institute highlighted a pessimistic perspective, demonstrating how a frontier AI model autonomously executed a 32-step cyber attack in a simulated environment. While research indicates that 'dark AI' tools have had limited impact on cyber crime due to a lack of user technical skills, experts warn that the uncertainty surrounding autonomous AI capabilities remains significant.
6K Energy and CRG Defense sign seven-year deal to support domestic battery supply for US defense systems
6K Energy and CRG Defense have announced a seven-year collaboration to establish a domestic supply chain for high-performance cathode active materials for US defense applications. The agreement, effective immediately, involves 6K Energy supplying single-crystal NMC811 cathode material from its North Andover, Massachusetts facility to CRG Defense. The partnership aims to reduce reliance on foreign supply chains in compliance with upcoming FCC bans and NDAA provisions regarding foreign-produced unmanned aerial system components. Additional capacity is expected from 6K Energy's Jackson, Tennessee facility starting in early 2028.