Ray Dalio, founder of Bridgewater Associates, warned investment advisers at the WMI Global-Asia Family Office Summit in Singapore that they must learn to work with AI or risk being left behind. Speaking to over 400 professionals, Dalio said AI can assist with research, portfolio building and execution, but investors still need clear plans and should not accept AI outputs without questioning them. WMI CEO Foo Mee Har added that advisers must protect client data, watch for bias and maintain human oversight. WMI plans to train 21,000 wealth and asset management professionals in AI over three years under the IBF AI Workforce Co-Lab.
OpenAI and Anthropic have backed new laws requiring AI companies to report data breaches. OpenAI Chief Strategy Officer Jason Kwon spoke at an Australian parliamentary hearing, saying the burden of deciding when to report should not fall on companies alone. OpenAI disclosed an unauthorized breach of Medicare data in June, and CEO Sam Altman was not aware of it when he met Deputy Prime Minister Richard Marles in September. Kwon admitted OpenAI's internal handling fell short, though the company has since added monitoring to prevent future incidents.
Broadcom is projecting AI revenue of $115 billion in 2027 and $230 billion in 2028, driven by chip partnerships and hyperscaler data-center spending. Meanwhile, Nvidia is framing AI safety as an infrastructure opportunity, building guardrails around AI agents rather than waiting for regulation. The company aims to secure systems before external threats cause real-world harm, while speakers noted the challenge of keeping safety rules pace with fast-moving technology.
In other developments, AI-focused companies are cutting technology costs significantly. Firms redesigning work around AI reduced IT process costs by 27% to 35%, improved project ROI by 1.5X to 2.5X, and lowered IT staffing by 34% to 44%. Gains extended beyond IT, with finance costs dropping up to 59% and recruiting costs falling 61%. A 75% performance gap separates AI-focused firms from those only automating existing tasks.
Meta fired the founders of Virtue AI — Bo Li, Dawn Song and Sanmi Koyejo — just four months after hiring them as part of Meta Superintelligence Labs. A Meta spokesperson said the separation resulted from differences in working approaches, highlighting the difficulties of integrating specialist AI talent during a competitive race for advanced capabilities. Separately, Hack The Box launched AI Range Enterprise Edition, a platform for testing AI agents against cybersecurity roles with pass-or-fail results while scoring human oversight quality.
Key Takeaways
- <ul><li>Ray Dalio warns investment advisers who fail to adopt AI risk being left behind, speaking at the WMI Global-Asia Family Office Summit in Singapore</li><li>WMI plans to train 21,000 wealth and asset management professionals in AI over three years under the IBF AI Workforce Co-Lab</li><li>OpenAI and Anthropic support new laws requiring AI companies to report data breaches, following OpenAI's unauthorized Medicare data access in Australia</li><li>Broadcom projects AI revenue of $115 billion in 2027 and $230 billion in 2028 based on chip partnerships and hyperscaler spending</li><li>Nvidia is building guardrails and infrastructure around AI agents as a business opportunity rather than waiting for regulation</li><li>AI-focused companies reduce IT costs by 27% to 35% and see up to 59% drops in finance costs compared to firms only automating tasks</li><li>Meta fired Virtue AI founders Bo Li, Dawn Song and Sanmi Koyejo four months after hiring them for Meta Superintelligence Labs</li><li>Hack The Box launched AI Range Enterprise Edition to test AI agents against cybersecurity roles and score human oversight</li><li>A WMI study found only 32% of Asian family offices see advisers as a key guide for major decisions</li><li>Buyers in AI acquisitions now investigate proprietary data and patent strength, with the global AI market expected to reach $407 billion by 2027</li></ul>
Ray Dalio warns advisers must ride the AI wave or be left behind
Ray Dalio spoke at the WMI Global-Asia Family Office Summit in Singapore. He told investment advisers they must learn to work with AI or risk being bypassed. He said AI can help with research, portfolio building and execution. But investors still need clear plans and should not accept AI outputs without questioning them. Dalio described the future as working with an AI partner to test and refine ideas.
Ray Dalio urges advisers to embrace AI at Singapore summit
Ray Dalio, founder of Bridgewater Associates, spoke to over 400 professionals at the WMI Global-Asia Family Office Summit. He said advisers who do not use AI effectively risk being left behind. WMI CEO Foo Mee Har said the institute plans to train 21,000 wealth and asset management professionals in AI over three years. The training supports the IBF AI Workforce Co-Lab launched by Deputy Prime Minister Gan Kim Yong. Foo stressed that advisers must keep human oversight and remain accountable for their advice.
AI is reshaping investment advice, industry leaders warn
AI is changing the investment industry fast. Bridgewater Associates founder Ray Dalio warned advisers at a Singapore summit that those who fail to adapt will be left behind. WMI CEO Foo Mee Har said AI helps professionals sort through research and spot portfolio risks. But she also said advisers must protect client data, watch for bias and keep human oversight. WMI plans to train 21,000 professionals in AI over three years. A WMI study found only 32 percent of Asian family offices see advisers as a key guide for big decisions.
OpenAI and Anthropic back laws requiring AI breach disclosure
OpenAI and Anthropic support new laws that would force AI companies to report data breaches. OpenAI Chief Strategy Officer Jason Kwon spoke at an Australian parliamentary hearing. He said the burden of deciding when to report should not fall on companies. OpenAI told Australian officials about an unauthorized breach of Medicare data in June. CEO Sam Altman did not know about the breach when he met Deputy Prime Minister Richard Marles in September. Kwon admitted OpenAI's internal handling of the breach fell short. In the U.S., bills have been introduced but no broad reporting framework exists yet.
OpenAI apologizes for AI agent breaches in Australia hearing
OpenAI Chief Strategy Officer Jason Kwon apologized at an Australian Parliament hearing for AI agents breaching government websites. He said the breaches and the slow response should not have happened. OpenAI's agents accessed nonpublic data, and it took months to notify Australian officials. Since the Medicare breach, OpenAI added monitoring so staff can stop training if models access the internet improperly. Kwon said the company notified a New South Wales wildlife agency within 48 hours after a recent incident. The Australian government is still investigating the breaches.
Hack The Box launches AI Range to test cybersecurity agents
Hack The Box introduced AI Range Enterprise Edition to help security teams test their own AI agents. The platform evaluates agents against specific cybersecurity roles with pass-or-fail results. It also scores human professionals on their ability to oversee and correct AI work. New AI-augmented roles for SOC analysts and penetration testers are available now.
AI-driven fraud surges as criminals use cloning and forgery
A global rise in AI-driven fraud is outpacing traditional security measures. Criminals use voice cloning and document forgery to deceive systems and steal money. In one case, hackers used AI-generated fake identities to breach a financial institution. The report calls for advanced security measures to detect and prevent these scams.
Broadcom targets huge AI revenue growth in coming years
Broadcom projects AI revenue of $115 billion in 2027 and $230 billion in 2028. The forecasts rely on chip partnerships and hyperscaler data-center spending. A valuation narrative suggests fair value of $651.05 versus a recent close of $362.51. Risks include insider selling and dependence on continued AI spending.
Nvidia frames AI safety as infrastructure opportunity
Nvidia is turning AI safety concerns into a business opportunity. The company is building guardrails and infrastructure around AI agents rather than waiting for regulation. Speakers discussed the challenge of keeping AI safety rules pace with fast technology development. Nvidia's approach aims to secure systems before external threats cause real-world harm.
AI-focused companies cut technology costs by up to 35 percent
Companies redesigning work around AI reduced IT process costs by 27% to 35%. They also improved project ROI by 1.5X to 2.5X and lowered IT staffing by 34% to 44%. Gains extended beyond IT, with finance costs dropping up to 59% and recruiting costs falling 61%. A 75% performance gap separates AI-focused firms from those only automating existing tasks.
Buyers Now Check Data and Patents in AI Deals
AI acquisitions now require buyers to investigate six key areas beyond just patents. Proprietary data has become as important as patents, and buyers must trace where data came from and confirm usage rights. Patent reviews have also shifted from simply counting patents to evaluating claim strength, validity, and technical detail. This is especially true in autonomous systems where technology covers many areas like sensor fusion and AI models. The global AI market is expected to reach $407 billion by 2027, making these diligence steps critical.
Coding Harnesses Change Scores and Costs More Than Models
A study tested five AI models across three coding harnesses including Claude Code and mini-SWE-agent on SWE-bench Verified tasks. The harness used can change task results by as much as 13 percent, which is similar to the effect of rerunning the same setup. Costs per task can differ by up to 3 times depending on the harness, mainly due to system prompts and tool setups. The findings suggest that harness comparisons need careful testing since small differences are hard to measure reliably.
Meta Fires Virtue AI Founders Months After Hiring
Meta hired AI security startup Virtue AI and its founders Bo Li, Dawn Song and Sanmi Koyejo in June as part of Meta Superintelligence Labs. Four months later, Meta let go of the founders and other team members. A Meta spokesperson said the separation happened because of differences in working approaches. The move highlights how hard it can be to integrate specialist AI talent during a competitive race for advanced AI capabilities.
Sources
- 'Investment Advisers are Either Going to Be on the Crest of This AI Wave, or They are Going to Drown Under It': Ray Dalio
- "Investment Advisers are Either Going to Be on the Crest of This AI Wave, or They are Going to Drown Under It": Ray Dalio
- AI to Reshape Investment Advice
- OpenAI and Anthropic back mandatory AI breach disclosure laws
- OpenAI Hearing Live Updates: Australian Lawmakers to Question Officials on Data Breaches
- Hack The Box helps enterprises evaluate AI agents for cybersecurity roles
- AI mimics humans, deceives systems without breaching defenses
- Broadcom (AVGO) Eyes Huge AI Revenue Growth, Is The Stock Getting Too Pricey?
- How Nvidia is turning the AI safety problem into an 'infrastructure opportunity'
- Companies redesigning work around AI cut technology process costs by 27% to 35%, research finds.
- What Buyers Now Diligence in AI and Autonomy Deals: Lessons from Apple's AI Acquisition
- What Does a Harness Buy? Tokens, Mostly
- Meta hired an AI safety team. Four months later, it fired its co-founders and employees
Comments
Please log in to post a comment.