LLMs hit security plateau: Why AI code can't be trusted yet

• Southeast Asia is experiencing an AI boom, driving significant investments in cloud infrastructure, semiconductor production, and data centers across the region. • Malaysia has become the fastest-growing data center market, benefiting from a 2019 moratorium on new facilities in Singapore due to land and energy constraints.

• The 48th global edition of the World AI Show took place in Kuala Lumpur on August 13, 2026, highlighting artificial intelligence as a central pillar of Malaysia's economic strategy. • The event focused on the country's accelerated investments in AI infrastructure and the development of sovereign cloud capabilities to bolster digital ambitions.

• In March, a supply chain attack targeting LiteLLM compromised over 2,500 companies via "Sandclock" infostealer malware. • The malware escalated privileges to root on build systems to steal cloud keys, SSH keys, Kubernetes secrets, and AI provider API keys.

• Credit Default Swaps (CDS) for major tech giants are triggering warnings as investors grow concerned over the sustainability of massive AI-related spending. • The surge in these financial instruments reflects market anxiety regarding the high levels of debt being accumulated to fund AI infrastructure and development.

• Hyperight has identified the top seven AI conferences scheduled to take place across Europe in 2026. • These events will prioritize critical technical and strategic topics, including agentic workflows, compute infrastructure, sovereign AI strategy, and compliance with the EU AI Act.

• A new field called "tokenomics" is emerging as companies struggle to accurately measure the return on investment (ROI) for their AI spending. • The Linux Foundation's Tokenomics Foundation is working to establish industry-wide standards to help firms evaluate the financial impact of AI investments.

• Mistral is shifting its business strategy from being solely an AI model developer to becoming a comprehensive AI infrastructure provider to better compete with global industry leaders. • In a significant move, the European lab will now host open-weight models from third-party developers on its platform, starting with the GLM-5.2 model from Chinese AI lab Z.ai.

• Venture capital firm Accel has raised $3.5 billion specifically to fund early-stage artificial intelligence startups. • The investment strategy targets innovative companies located across three primary global regions: the United States, Europe, and India.

• AI security startup Mindgard has successfully closed a $30 million Series A funding round to address vulnerabilities in AI attack surfaces. • This latest investment brings the company's total funding to nearly $42 million, signaling strong investor confidence in AI-specific defense tools.

• Lancaster University spinout Mindgard has raised $30 million to scale its AI security platform after successfully bypassing guardrails in ChatGPT, Cursor, and Google's AI tools. • The funding will be used to accelerate growth across product development, engineering, sales, and marketing to address critical vulnerabilities in large language models.

• CrowdStrike's 2026 Threat Hunting Report has identified a new cyberattack trend called "LLMJacking," where hackers hijack corporate AI access to inflate usage costs. • In one highlighted incident, attackers utilized a stolen cloud identity to execute nearly 200,000 API requests within a two-minute window.