B2BROKER has expanded its B2TRADER platform with AI trading agents and in-terminal chat, giving traders the ability to issue natural-language commands to close positions or set stop-losses. Brokers can choose between read-only access for analysis or full access for automated trading. CEO Arthur Azizov said brokers retain control over the boundaries of what the AI tools can do. The update builds on earlier embedded assistants that offered market forecasts and trading signals.
Anthropic is preparing for a Wall Street debut that could reshape how private AI companies are valued. The company behind Claude has filed for an IPO on the Nasdaq under the ticker ANT, with a potential debut as early as November 16. Anthropic reported nearly 12,000 customers with revenue over $100,000 as of September 30, and has committed at least $855 billion to compute and infrastructure. Major backers include Amazon, Spark Capital, and Fidelity, with private investors previously valuing the company at up to $2 trillion.
OpenAI has built an autonomous pipeline that found and patched 948 local privilege-escalation vulnerabilities in the Linux kernel. Scout agents identified high-value audit targets, which Analyst agents then scanned. The pipeline reviewed roughly 2.8 million lines of code, about 7% of the kernel, and creates proof-of-concept exploits, candidate patches, and root cause analyses. A human security researcher still reviews all results before fixes are shared publicly.
On the security front, Meta's AI agent Muse drew sharp criticism from former Meta security chief Alex Stamos, who called the agent's access to personal messages, email inboxes, and sensitive data like passport numbers "totally insane." Security researchers have found vulnerabilities that could expose personal information. Separately, a Barracuda Research survey of 2,000 IT and security leaders found 66% of organizations already work with a managed service provider, and about 59% worry about AI-related risks such as AI-generated phishing.
In telecom and infrastructure, Chief Telecom is building the Huwei AI Data Center in Yunlin County, Taiwan, with a total investment of TWD 3.3 billion and a capacity of 16 MW, expected to begin commercial operation in the second half of 2028. Lumen Technologies, Telus, and Etiya jointly stated that AI is transforming how telecom vendors operate, helping build networks, prevent outages, and improve customer experiences. Meanwhile, banks are testing AI agents with strict application gates, using golden datasets and deterministic tools to push numerical accuracy from 60% to 95%.
Key Takeaways
- B2BROKER added MCP trading agents and in-terminal chat to B2TRADER, allowing natural-language trading commands with broker-controlled access levels.
- Anthropic's IPO could debut on Nasdaq as ANT as early as November 16, with over $855 billion committed to compute and infrastructure.
- Amazon is a major backer of Anthropic, which reports nearly 12,000 customers with revenue over $100,000.
- OpenAI's autonomous pipeline patched 948 Linux kernel vulnerabilities across roughly 2.8 million lines of code, with human review before public fixes.
- Former Meta security chief Alex Stamos criticized the Muse AI agent for accessing personal messages, emails, and sensitive data like passport numbers.
- Chief Telecom is investing TWD 3.3 billion in a 16 MW AI data center in Taiwan, expected online in the second half of 2028.
- Banks using AI agents with strict application gates improved numerical accuracy from 60% to 95% via golden datasets and deterministic tools.
- 66% of organizations work with managed service providers, and 59% worry about AI-related risks like AI-generated phishing, per a Barracuda survey.
- Lumen, Telus, and Etiya say AI is transforming telecom vendor relationships, network operations, and customer experience.
- KierstonBench tested 16 language models on crime investigation tasks in a simulated city of 125,000 residents, finding most models could not perform records investigation reliably.
B2BROKER Adds AI Trading Agents and Chat to B2TRADER
B2BROKER expanded its B2TRADER AI strategy with MCP trading agents and in-terminal chat. Traders can give agents natural-language commands to close positions or set stop-losses. Brokers choose between read-only access for analysis or full access for automated trading. The in-terminal chat lets traders ask about markets and positions directly inside the B2TRADER web terminal. CEO Arthur Azizov said brokers control the boundaries of what AI tools can do.
B2BROKER Expands AI Features for B2TRADER Platform
B2BROKER Group announced it expanded its B2TRADER AI strategy with MCP trading agents and in-terminal chat. The new features aim to give clients a more personalized and efficient trading experience. Clients can access a wider range of trading tools through the B2TRADER platform. The company said the update builds on earlier AI tools like an embedded assistant offering market forecasts and trading signals.
B2BROKER Enhances B2TRADER With AI Trading Tools
B2BROKER Group expanded its B2TRADER AI strategy by adding MCP trading agents and in-terminal chat. The update is designed to provide traders with real-time market data and insights. B2BROKER aims to improve its services and offer a more comprehensive trading experience. The company is a global fintech solutions provider based in Dubai.
Anthropic IPO to Influence How Private AI Stocks Are Valued
Anthropic's upcoming Wall Street IPO will bring new price transparency to private AI companies. Public data from Anthropic will set a higher standard for judging other AI labs like OpenAI, Mistral, and Cohere. Anthropic has nearly 12,000 customers with revenue over $100,000 as of September 30. The company has committed at least $855 billion to compute and infrastructure. Private investors previously valued Anthropic's stock at up to $2 trillion.
What to Know Before Anthropic's IPO Debut
Anthropic, the AI lab behind Claude, has filed to go public and may debut as early as November 16. The stock ticker symbol will be ANT on the Nasdaq Global Select Market. Anthropic's S-1 filing revealed commitments to spend $895.4 billion on compute and infrastructure. Major backers include Amazon, Spark Capital, and Fidelity. The company has not yet announced the number of shares or price range for its offering.
DualWorldBench tests AI agents across symbolic and grounded worlds
Researchers introduced DualWorldBench, a benchmark for testing multimodal AI agents on complex planning tasks. It includes TextWorld for symbolic planning, SimWorld for grounded path planning, and DualWorld-VQA for perception and reasoning diagnostics. Experiments show agents struggle to combine local understanding with global planning. All datasets and code are publicly available.
Chief Telecom invests billions in AI data center in Taiwan
Chief Telecom is building the Huwei AI Data Center in Yunlin County, Taiwan. The project will require a total investment of TWD 3.3 billion. The facility will have a capacity of 16 MW. Commercial operation is expected to begin in the second half of 2028.
Banks test AI agents with strict application gates
A bank told AI startup teams it certifies applications, not models. The team used a golden dataset and deterministic tools to improve numerical accuracy from 60% to 95%. Domain experts review failures and update the dataset. Any change to the system prompt, model, or tool triggers a new evaluation run.
Former Meta security chief criticizes Muse AI agent privacy risks
Meta's AI agent Muse can access personal messages, email inboxes, and sensitive data like passport numbers. Former Meta security chief Alex Stamos called the approach totally insane. Security researchers have found vulnerabilities that could expose personal data. The article also notes concerns about AI startups bypassing bot detection systems.
Barracuda Research says AI security will drive MSP growth
A survey of 2,000 IT and security leaders found 66% of organizations already work with a managed service provider. Many companies seek help because they cannot respond to threats quickly. About 59% worry about AI-related risks like AI-generated phishing. Organizations also value providers who explain risks in plain language.
Lumen, Telus and Etiya say AI is transforming telecom
Lumen Technologies, Telus, and Etiya see AI as a major force in telecom. A new report shows AI is changing how vendors work in the industry. Companies are using AI to build networks, prevent outages, and improve customer experiences. AI also helps automate tasks and spot problems before they grow. The report says traditional vendor relationships and business models will need to change as AI use grows.
OpenAI pipeline finds and patches Linux kernel flaws
OpenAI built an autonomous pipeline that found and patched 948 local privilege-escalation vulnerabilities in the Linux kernel. Scout agents identified high-value audit targets, which Analyst agents then scanned for vulnerabilities. The pipeline reviewed about 5,034 audit targets covering roughly 2.8 million lines of code, or about 7% of the Linux kernel. It creates proof-of-concept exploits, candidate patches, and root cause analyses to help human researchers. A human security researcher still reviews all results before fixes are shared.
Kierston tests language models on crime investigation tasks
Kierston is a city-scale simulation engine modeling about 125,000 residents of Allentown, Pennsylvania. KierstonBench is a benchmark that tests language models on crime investigation and surveillance tasks using simulated records. The benchmark includes 32 task families such as record linkage and incident investigation. In tests of sixteen language models, most could not perform records investigation and made unsupported identifications. A few frontier models solved most tasks but still sometimes fell for deceptions and did not replace human investigators.
Sources
- B2BROKER Expands B2TRADER AI Strategy With MCP Trading Agents and In-Terminal Chat
- B2BROKER Expands B2TRADER AI Strategy With MCP Trading Agents and In-Terminal Chat
- B2BROKER Expands B2TRADER AI Strategy With MCP Trading Agents and In-Terminal Chat
- Anthropic’s IPO to Cast Long Shadow Over Private AI Stocks
- Everything you need to know ahead of Anthropic's IPO
- DualWorldBench: Can Agents Plan Deliveries across Symbolic and Grounded Worlds?
- Chief Telecom to invest TWD 3.3 billion in 16 MW AI data centre in Taiwan
- Banks Don't Certify Models. They Gate Apps.
- Former Meta Exec Shreds the Company's Muse AI Agent as "Insane" Privacy Risk
- Barracuda Research: AI Security and Cyber Resilience Will Drive MSP Growth
- AI is rewriting telecom's old vendor playbook
- A thousand fewer paths to root
- Kierston & KierstonBench: Simulation of Allentown, PA and Evaluation of Language Models in Crime Investigation and Surveillance Tasks
Comments
Please log in to post a comment.