Malaysia’s Malaysia Digital 2030 plan aims to make the country an AI nation by 2030, targeting a digital economy worth 30% of GDP, 500,000 high-value digital jobs, and more locally developed digital products and intellectual property. Digital Minister Gobind Singh Deo said the AI Action Plan for 2026–2030 has no separate budget and will draw on existing public funds, public-private partnerships, and local and foreign investment. ICT and e-commerce contributed RM476.8 billion, or 23.5% of GDP, in 2025.
In India, Karnataka plans to train 63,200 students and working professionals with Google, Salesforce, Pega Systems, Sarvam AI, and Gnani.ai. Announced by Chief Minister DK Shivakumar on November 17 at the Bengaluru Tech Summit, the programme covers generative and agentic AI, voice-based AI, cloud computing, data analytics, and enterprise platforms, with projects and certification opportunities. Saudi AI company Humain has also announced Humain+, a hybrid skills programme with tracks for leaders, non-technical professionals, technical talent, and students. Its Riyadh learning center is expected to open in the first quarter of 2027.
Crypto security executives warn that AI is making attacks easier to develop and execute. Anchorage Digital co-founder Diogo MĂłnica and Solana Foundation security chief Michael Coates cited automated vulnerability discovery and exploit creation, alongside North Korean tactics such as fake job interviews, poisoned tools, and deepfake calls. They urged companies to avoid basic security lapses, including giving AI agents access to email and failing to protect keys with hardware security keys. Separately, OpenAI says a DNS configuration error gave a research model limited internet access. The company paused its most capable training from September 20 to October 1, and says its Control Room can stop output within 35 seconds of a halt order.
A study of five large language models from four vendors found none beat simple trading rules after costs when tested around 598 earnings announcements and 409 bad-news drops in large US stocks. A GPT-based model correctly predicted 59.2% of announcement directions before its training cutoff, but 53.3% after it; researchers said the only gains came from information memorized during training. DigitalOcean, meanwhile, reported 40% year-over-year Q3 revenue growth and US$234 million in annual recurring revenue from AI customers. New offerings include the Cuttlefish iPhone app for Managed Agents, a Box content integration, and Cloudways AI Website Builder.
Strike Graph is using AI agents to create read-only integrations that collect compliance data through APIs. CEO Justin Beals said the system checks documentation and avoids handling customer credentials; people review the plan, code, and deployment, and evidence scripts run in isolated sandboxes. Researchers have also proposed a multi-agent framework for food assistance that prompts systems to seek clarification when information is uncertain and keeps assumptions consistent as tasks pass between agents.
The Federal Reserve’s November 2026 Financial Stability Report says 87% of surveyed market contacts identified AI as a top financial risk, as AI enthusiasm partly supports high stock valuations. The report’s practical investing guidance includes checking overlapping holdings, aligning stock exposure with when funds are needed, and rebalancing on a schedule; redirecting new contributions may avoid capital-gains taxes in taxable accounts. In market news, Anthropic shares were indicated to open at $99, implying a fully diluted valuation of $1.7 trillion, after an IPO priced at $86 and marketing around $75. Nvidia rose 4% and Micron 6% as the Nasdaq gained 1.9% early in trading.
Nvidia also faces a California investigation. Attorney General Rob Bonta served the company with an investigative subpoena as he examines possible transfers of advanced chips or confidential information to China and whether Nvidia or CEO Jensen Huang misled investors or the public. The inquiry follows reports involving Megaspeed, a Singapore-based data center company, and concerns that Nvidia chips may have been diverted to China. Nvidia shares have fallen 22% since a New York Times report in early November; board chair Harvey Jones said the company does not plan to reinstate Huang, who left last week.
Key Takeaways
- Malaysia’s 2030 targets include a digital economy equal to 30% of GDP and 500,000 high-value digital jobs; its AI action plan relies on existing funds and private investment.
- Karnataka will train 63,200 people with Google, Salesforce, Pega Systems, Sarvam AI, and Gnani.ai in AI and other technology areas.
- Humain+ will provide four AI learning tracks in Saudi Arabia, with its Riyadh center expected to open in Q1 2027.
- Crypto security leaders warn AI lowers the cost of attacks and highlight North Korean tactics; they advise against exposing email to AI agents and stress hardware security keys.
- OpenAI paused its most capable training from September 20 to October 1 after a DNS error gave a research model limited internet access.
- A study found five LLM trading agents did not beat simple strategies after costs; a GPT-based model’s directional accuracy fell from 59.2% before its training cutoff to 53.3% after.
- DigitalOcean reported Q3 revenue growth of 40% and US$234 million in AI customer ARR, alongside new agent, Box, and website-builder offerings.
- Strike Graph’s AI-built compliance integrations require human review and use secure credential handling and isolated evidence-script sandboxes; a separate research framework helps food-assistance agents clarify uncertain information.
- The Fed says 87% of surveyed market contacts see AI as a top financial risk; its report also notes elevated stock valuations and offers guidance on managing portfolio exposure.
- Anthropic shares were indicated at $99, implying a $1.7 trillion valuation, while Nvidia faces a California subpoena over possible China chip transfers and related disclosures.
Malaysia bets on investment to build a homegrown AI industry
Malaysia aims to become an AI nation by 2030 through its Malaysia Digital 2030 plan, which includes the AI Action Plan 2026 to 2030. Digital Minister Gobind Singh Deo says the plan has no separate budget and will rely on existing government funds, public private partnerships, and foreign and local private investment. Its goals include raising the digital economy’s share of GDP to 30%, creating 500,000 high value digital jobs, and supporting Malaysian digital products and intellectual property by 2030. ICT and e-commerce contributed RM476.8 billion, or 23.5% of GDP, in 2025, but the article says Malaysia still needs to turn infrastructure investment into local AI skills, companies, and products.
Karnataka to train 63,200 people in AI with five tech firms
Karnataka will train 63,200 students and working professionals in AI and other emerging technologies with Google, Salesforce, Pega Systems, Sarvam AI, and Gnani.ai. Chief Minister DK Shivakumar announced the programme on November 17 at the opening of the 29th Bengaluru Tech Summit 2026. Training will cover generative AI, agentic AI, voice based AI, cloud computing, data analytics, and enterprise platforms, with practical projects and certification opportunities. The three day summit runs through November 19 at the Bengaluru International Exhibition Centre under the theme “AI and Beyond.”
Humain+ offers hands on AI training across Saudi Arabia
Saudi AI company Humain has introduced Humain+, a hybrid learning programme designed to build practical AI skills for students, workers, and business leaders. It will offer four learning tracks for leaders, non technical professionals, technical talent, and students, covering topics such as AI strategy, governance, product development, agentic AI, security, and infrastructure operations. The programme will combine a learning center in Riyadh with online and distance learning, and will include courses tailored to industries such as health care, finance, and tourism. The center is expected to open in the first quarter of 2027.
Crypto security leaders warn AI attacks could surge over the next year
Crypto security leaders warned that AI could make attacks against crypto companies more frequent and more powerful over the coming year. Anchorage Digital co founder Diogo MĂłnica said AI systems can now find weaknesses, build exploits, and carry out attacks, while Solana Foundation security chief Michael Coates said the tools have lowered the cost and difficulty of hacking. Coates identified North Korea as a major threat, citing fake job interviews, poisoned tools, and deepfake video calls, including tactics linked to the Binance attack. Both executives warned companies against basic security mistakes, including connecting email inboxes to AI agents and failing to protect keys with hardware security keys.
Crypto security leaders warn of AI powered attacks and North Korean threats
At Solana’s Breakpoint conference in London, security leaders warned that crypto companies face a wave of AI powered attacks over the coming year, alongside continued threats from North Korea. The article identifies Anchorage Digital co founder Diogo Mónica and Solana Foundation security chief Michael Coates as speakers in the warning.
OpenAI pauses GPT-6.1 Astra training after containment issue
OpenAI says frontier models such as GPT-6.1 Astra train in restricted virtual machines and cloud networks without live internet access. The company uses monitoring and alignment checks to help catch model errors or attempts to bypass limits. OpenAI says its Control Room can stop output within 35 seconds of a halt order. After monitors found a research model had gained limited internet access through a DNS setup error, OpenAI paused its most capable training from September 20 to October 1.
Study finds LLM trading agents fail to beat simple earnings bets
Researchers tested five large language models from four vendors on trades before 598 earnings announcements and after 409 bad-news drops among large US stocks. The benchmark used only information available by each decision close and compared agents with simple rules and logistic regression. No agent beat the simple bets after costs, and a Gemini-based agent did not pick better dips than buying every drop. The only gains came from information models had memorized during training. A GPT-based model predicted the direction of 59.2% of announcements before its training cutoff and 53.3% after it.
DigitalOcean grows AI business with new tools and integrations
DigitalOcean reported its fastest growth in a decade, with Q3 revenue up 40% and AI customer annual recurring revenue reaching US$234 million. New AI-focused products include the Cuttlefish iPhone app for Managed Agents, a Box content integration, and Cloudways AI Website Builder. These tools expand how customers deploy and manage AI workloads. The article says DigitalOcean is moving beyond basic infrastructure into AI workflows linked to its inference cloud and core services.
Strike Graph uses AI agents to build secure GRC integrations
Strike Graph added a feature that uses AI agents to build read-only integrations for collecting compliance data through APIs. CEO Justin Beals said the system checks API documentation, avoids handling customer credentials, and stores sensitive data in a secure vault. Human reviewers must check the AI's plan, its code, and the final deployment, while a separate locked-down process publishes integrations. Once deployed, evidence scripts run in isolated sandboxes. Beals said AI-built connectors could help GRC platforms keep pace as organizations add more applications.
New AI framework improves clarification in food assistance
Researchers introduced a framework for multi-agent food assistance systems that helps them ask for clarification when information is uncertain. It keeps shared assumptions consistent as tasks pass between agents, reducing the risk that tentative interpretations become treated as facts. Each agent identifies what decisions it can make and what evidence it still needs. The framework assesses information questions and user-approved fallback options together, while accounting for follow-up interactions and shared limits. It aims to support food resource navigation, nutrition guidance, and supply planning.
Fed flags AI risks and offers ways to check retirement exposure
The Federal Reserve’s November 2026 Financial Stability Report found that 87% of surveyed market contacts named artificial intelligence as a top financial risk. High stock values are partly driven by enthusiasm for AI, and the S&P 500’s forward price-to-earnings ratio remains above its historical median. Investors can check the top 10 holdings in their index funds, look for overlap across accounts, match stock exposure to when they need the money, and rebalance on a schedule. Redirecting new contributions instead of selling investments may help avoid a capital gains tax bill in taxable accounts.
Anthropic IPO draws strong demand as shares point to a $1.7 trillion value
Anthropic shares were indicated to open at $99 each, which would value the company at $1.7 trillion on a fully diluted basis. The IPO price was $86, up from about $75 during marketing, and a report said institutional buyers were shut out. The Nasdaq rose 1.9% early in trading, with Nvidia up 4% and Micron up 6%. The report says investors will be watching the market’s response at the close and on Friday, while some worry that the record IPO could signal a market top.
California investigates Nvidia over possible China chip transfers
California Attorney General Rob Bonta has served Nvidia with an investigative subpoena. He is examining whether Nvidia or its CEO, Jensen Huang, helped China obtain advanced chips or confidential information, and whether the company misled investors or the public in violation of state law. The probe follows reports about Huang’s relationship with Alice Huang, an executive at Singapore-based data center company Megaspeed, and concerns that Megaspeed may have diverted Nvidia chips to China. Nvidia shares have fallen 22% since a New York Times report in early November, and board chair Harvey Jones said the company does not plan to reinstate Huang, who left last week.
Sources
- Can Malaysia build its own AI industry beyond data centres?
- ​Karnataka Announces AI Training Programme for 63,200 Students and Professionals
- Humain targets Saudi Arabia’s AI skills gap with hands-on training
- AI Will Make the Next 12 Months 'a Slaughterhouse' for Crypto Security: Anchorage Exec
- AI Will Make the Next 12 Months 'a Slaughterhouse' for Crypto Security: Anchorage Exec
- OpenAI slows models to keep containment
- The Only Edge Was Memorized in Training: LLM Trading Agents Around Earnings and Bad News
- Will AI Product Adoption Change DigitalOcean Stock Narrative
- Strike Graph Adds AI Agents to Build GRC Platform Integrations
- Receiver-Conditioned Clarification for Multi-Agent Food Assistance
- AI Just Topped the Fed’s List of Financial Risks. 4 Ways to See How Much of Your Retirement Rides on It
- Strong Anthropic IPO gives new life to the AI trade. Shares indicated to open at $99
- California probing whether China improperly got Nvidia technology
Comments
Please log in to post a comment.