Amazon Calls for Mandatory AI Safety Standards

Key Takeaways

• <ul><li>Amazon calls for AI models to be rigorously tested before release, aligning with OpenAI, Anthropic, Google, Microsoft, and others on collective safety standards</li><li>Nvidia CEO Jensen Huang forecasts chip sales will double by 2027, reflecting sustained AI infrastructure demand</li><li>Global data center investment could reach $50 trillion by 2050, with the US accounting for $15.1 trillion of that total</li><li>Doosan Group invests ₩970 billion (~$701 million) to expand copper clad laminate production for AI data centers through 2028</li><li>Credo Technology Group revenue tripled to $1.335 billion in fiscal 2026 on strong AI data center component demand</li><li>Druva launches AI-powered Ransomware Detection and Metagraph tool that cuts attack investigation time from days to hours</li><li>Pilot releases Navigator, an AI tool for hospitality that analyzes sales trends, supplier prices, and menu profitability</li><li>Advanced AI Society and Linux Foundation launch Proof-of-Control v1.0 for continuous, inspectable AI agent verification</li><li>Amazon develops AI through AWS, an AGI unit powering Alexa, and the Bedrock marketplace for model access</li><li>Digital twin research using Melbourne data shows physics-constrained methods improve detection of false data in pedestrian sensor systems</li></ul>

Amazon says AI models should only be released when safe

Amazon stated that AI models must be fully tested and safe before being released to the public. Spokesperson Tom Parnell said innovation and security should not be seen as opposing goals. The company wants tech firms and government regulators to work together on mandatory safety standards. This stance aligns Amazon with OpenAI and Anthropic. It comes amid growing concerns about AI risks in Silicon Valley.

Amazon calls for rigorous AI testing and safeguards

Amazon entered the AI safety debate by saying models should only be released when rigorously tested and proven safe. The company sent its views in a letter to the Future of Life Institute. This move came as the tech industry faces growing concerns about risks from AI technologies like self-driving cars and facial recognition. Amazon's position marks a significant step in the ongoing discussion about AI oversight.

Amazon urges careful AI development without slowing progress

Amazon said AI models should be released only when they are ready and safe, based on rigorous testing and strong safeguards. A spokesperson told Reuters that progress and safety are not mutually exclusive. The company did not call for an industry slowdown but warned of risks if safety measures are not applied collectively. Amazon develops AI through AWS and an AGI unit that powers Alexa. It also provides infrastructure and model access through its Bedrock marketplace.

Amazon urges partnership on AI safeguards with government

Amazon stated that AI models should be released only when they are ready and safe through rigorous testing. The company said there are risks if AI development does not happen collectively. Amazon believes the industry should work with government to establish proper protections. This came after several large AI labs called for a slowdown in AI development. The concern followed incidents where AI agents conducted unauthorized cybersecurity attacks.

Amazon speaks out on AI safety after industry-wide concerns

Amazon weighed in on the AI safety debate, saying models should be safe and fully tested before release. The company did not call for an industry slowdown but stressed collective action with government. Amazon develops AI through AWS and an AGI unit for services like Alexa. It also operates the Bedrock marketplace for model access. The statement followed calls from OpenAI, Google DeepMind, xAI, and Microsoft for a more measured approach to AI development.

Doosan to invest ₩970 billion in AI CCL capacity expansion

Doosan Group is investing ₩970 billion (about $701 million) to expand production of copper clad laminate, a key material for AI data centers. ₩420 billion will expand CCL lines in South Korea and ₩550 billion will build a new plant in China. The investment will roll out through 2028. The company also approved canceling about 2.56 million treasury shares worth roughly ₩3 trillion.

AI Data Centers to attract $50 trillion in investment by 2050

A Samil PwC report says global data center investment could reach $50 trillion by 2050 if AI adoption accelerates. ICT equipment like GPUs and servers will make up about 93% of total spending because equipment must be replaced every 4 to 6 years. Power availability is the most important factor for investment decisions. The Americas are expected to capture 48% of total investment, with the United States accounting for $15.1 trillion of that.

Nvidia CEO Jensen Huang forecasts doubling chip sales in 2027

Nvidia CEO Jensen Huang told investors the company will sell twice as many chips in 2027. The forecast signals that the AI infrastructure boom is continuing strong. Huang believes the shift toward AI-first computing across industries is driving this growth. The prediction has drawn close attention from the semiconductor industry and investors.

Nvidia CEO forecasts 2x chip sales growth for next year

Nvidia CEO Jensen Huang predicted the company will sell double its chips next year, sending ripples through tech markets. The forecast highlights explosive demand for AI infrastructure that shows no signs of slowing. Enterprise customers are racing to build AI capabilities, and cloud providers are expanding GPU clusters. Analysts are recalibrating their models based on Huang's guidance, though the doubling must translate into real business value to sustain growth.

Digital twins detect false data in urban pedestrian sensors

Researchers developed physics-constrained digital twins to protect pedestrian counting systems from false data injection attacks. Using six years of Melbourne data, the method achieved a 0.54 attack margin against a single compromised device. The margin dropped to 0.19 when a third of sensors were compromised. The study shows that using a street graph with flow conservation laws provides far better detection than a simple distance graph.

Pilot launches AI tool Navigator for hospitality industry

Pilot, a point of sale platform from Paycorp Group, launched an AI layer called Navigator. It can spot sales trends, compare supplier prices, flag purchasing issues and identify less profitable menu items. The launch comes as competition grows in the hospitality sector. Company leader Glenn Miller says operators need actionable insights, not more reports.

AI agents depend on strong data foundations

AI agents need high quality data to work well because they act on behalf of companies. Data is often fragmented across different systems, making it hard for AI to use it effectively. Causal AI can map cause and effect but needs a continuous feedback loop. Clean and unified data streams help AI learn and improve over time.

Credo Technology sees AI revenue growth but stock lags

Credo Technology Group, which supplies AI data center components, saw revenue triple to $1.335 billion in fiscal 2026. Growth was driven by strong demand for optical and interconnect products. The stock is down 8.29% over the past year and trades at a forward P/E of about 35x. Analysts have a consensus target near $281, while long term hopes include reaching $400 by 2030.

Advanced AI Society launches open verification system

The Advanced AI Society joined the Linux Foundation and launched Proof-of-Control v1.0, an open verification ecosystem for AI agents. It aims to provide continuous monitoring that anyone can inspect, rather than relying on periodic audits. The group hosted an event called Who's Watching the Machines? with the Linux Foundation. Supporters include Charles Iheagwara, Bruce Schneier and J. Christopher.

Xtend AI Robotics stock surges 63% raising valuation questions

Xtend AI Robotics, a defense focused AI hardware and software company, has gained 62.8% over the past year. Investors are questioning whether the current stock price matches the company's actual sales. The company's AI powered defense solutions are gaining traction. The key concern is whether the valuation is supported by real revenue growth.

Druva uses AI to detect and recover from AI-powered ransomware attacks

Druva announced new AI-powered tools to fight ransomware attacks that also use AI. The new Ransomware Detection system uses AI and machine learning to find ransomware behavior across backup snapshots. It validates threats using forensics and turns evidence into recovery actions through Recovery Insights. Druva also introduced Dru Metagraph, which maps how attacks spread across identities and systems, cutting investigation time from days to hours. This helps security teams understand attacker behavior and create precise recovery plans.

Sources

• AI Security Debate — Amazon Says AI Models To Be Released Only When 'Ready And Safe'
• Amazon enters AI safety fray, calls for "rigorous testing," safeguards
• Amazon enters AI safety fray, calls for “rigorous testing,” safeguards
• Amazon Urges Industry and Government Partnership on AI Safeguards
• Amazon enters AI safety fray, calls for "rigorous testing," safeguards
• Doosan to Invest ₩970 Billion in AI CCL Capacity Expansion; Announces ₩3 Trillion Share Buyback and Cancellation
• AI Data Centers to Draw $50 Trillion by 2050, With Investment Shifting to Power and Equipment
• Nvidia CEO Jensen Huang Forecasts Doubling Chip Sales in 2027
• Nvidia CEO Forecasts 2x Chip Sales Growth for Next Year
• Physics-Constrained Digital Twins for Sensor Integrity in Urban Pedestrian Flow: Detecting Stealthy False Data Injection with Conformal Guarantees
• Pilot launches AI layer as competition in hospitality intensifies
• AI Agents Are Only As Good As The Foundation They Run On
• Credo Technology's AI infrastructure sales soar but stock lags; $400 target by 2030 is possible with growth and capex support.
• Advanced AI Society joins the Linux Foundation, launches open verification ecosystem as Congress moves on agent security
• Is Xtend AI Robotics (XTND) Above Fair Value After A 63% Run?
• Druva reacts to AI ransomware raids with AI responses