OpenAI Codex updates as Google Calendar integrates

Key Takeaways

• OpenAI's Codex AI received major updates, enabling it to operate computers, debug with Xcode, generate images, and integrate with over 90 developer tools like Google Calendar and GitHub.
• Aldo d'Aponte, CEO of Arbitrage Group Properties, pleaded guilty to using AI to generate fake complaint letters against London's Heaven nightclub, identified by an AI detection tool.
• AI is increasing risks for executive protection by lowering the barrier for attackers to create personalized phishing and doxing campaigns.
• Worker resistance to AI tools is significant, with over 80% of employees avoiding or rejecting them, though AI is expected to become essential for financial management.
• AI startup Speak's CTO, Andrew Hsu, is shifting hiring to prioritize an "agentic mindset" and proficiency with AI tools like Claude Code or Codex over traditional coding skills.
• Erica Brescia of Redpoint Ventures advocates for responsible and open-source AI development in cybersecurity, highlighting the need for public-private collaboration.
• GitLab introduced new AI features, including Agentic SAST Vulnerability Resolution for automatic code flaw fixes and agents in its Duo Agent Platform for CI pipeline setup.
• Nir Zuk, founder of Palo Alto Networks, is seeking to acquire voting shares in Liberty Bank's holding company, potentially to test AI tools in financial services.
• Canva launched Canva AI 2.0, allowing its 265 million monthly users to generate designs using conversational language, aiming to protect its $42 billion valuation.
• Globant expanded its collaboration with Autodesk to implement digital twin technology, combining its AI expertise with Autodesk Tandem for operational efficiency.

OpenAI Codex enhances coding with Xcode, image generation

OpenAI's Codex AI is becoming a more powerful tool for developers. It can now work with Xcode to debug applications and generate images. Codex also connects to many developer tools to manage workflows. This makes it easier to test apps, find bugs, and create assets like images for websites. The AI can even manage project files and integrate with services like Google Calendar and GitHub.

OpenAI Codex expands capabilities for developers

OpenAI has updated its Codex AI model to be a more comprehensive assistant for developers. Codex can now operate your computer, use more tools, generate images, and remember your preferences. It also supports developer workflows like reviewing code changes and connecting to remote development environments. The update includes an in-app browser for faster iteration on designs and apps. Over 90 new plugins enhance its ability to gather context and take action across various tools.

OpenAI Codex AI gets major update for developers

OpenAI has released a major update for its Codex AI model, enhancing its capabilities for millions of developers. The new version allows Codex to operate computer applications, generate images, and remember user preferences. It offers deeper support for the software development lifecycle, including handling code review comments and managing multiple terminal tabs. Codex can now also carry work forward over time by reusing conversation threads and proactively suggesting tasks.

Man used AI to fake complaints against London nightclub

A businessman named Aldo d'Aponte pleaded guilty to making false statements to try and shut down London's Heaven nightclub. Police believe he used AI to generate letters that appeared to be from neighbors. These letters contained detailed complaints about the club's reopening. An AI detection tool flagged the letters as likely AI-generated, and police traced the IP addresses to d'Aponte. He was given a conditional discharge and ordered to pay costs.

Man used AI to fake complaints against London nightclub

Aldo d'Aponte, CEO of Arbitrage Group Properties, pleaded guilty to submitting fake objection letters to shut down London's Heaven nightclub. Police suspect these letters were created using AI. The letters, sent from an encrypted email, were detailed complaints about the club. An AI detection tool identified them as likely AI-generated, and police traced the origin to d'Aponte. He received a 12-month conditional discharge and must pay costs.

Workers resist AI tools despite growing use

A new survey shows that over 54% of workers bypassed company AI tools in the past month, choosing to do work manually. Another 33% haven't used AI at all, meaning about 80% of employees are avoiding or rejecting the technology. Despite this resistance, many people expect AI to become essential, especially in managing finances. While some worry about losing human interaction, others see AI as a way to unlock new financial opportunities, particularly among younger generations.

AI creates new risks for executive protection

AI is making it easier for attackers to target corporate executives with personalized phishing emails and doxing campaigns. Previously, these attacks required technical skills and access to specialized data, but AI has lowered the barrier to entry. This means even amateurs can craft convincing messages that mimic company news or executive tones. Security teams must now use AI defensively to identify and fix exposed information before attackers can exploit it, protecting both executives and the organization.

Palo Alto Networks founder seeks stake in California bank

Nir Zuk, the founder of cybersecurity firm Palo Alto Networks, has applied to acquire voting shares in Liberty Bank's holding company, DMG Bancshares. While it's unclear if he plans a majority takeover, observers suggest the bank could become a test site for AI tools in financial services. Zuk, who co-founded an AI platform called eOS, may use this investment to demonstrate AI's capabilities in banking. This move aligns with a trend of tech leaders investing in financial services.

Erica Brescia: Responsible AI development is key for cybersecurity

Erica Brescia of Redpoint Ventures emphasizes the need for responsible development of AI models in cybersecurity. She notes that while AI can be used for attacks, it's also crucial for defense. Brescia advocates for open-source AI, believing it helps researchers find and fix vulnerabilities faster. She also highlights that successful companies are building their own AI tools to manage financial operations and other tasks, stressing the importance of collaboration between private and public sectors to ensure safety.

AI startup CTO changes hiring to focus on agentic skills

Andrew Hsu, CTO of AI startup Speak, is changing how they hire engineers. Traditional coding tests are gone because AI models can answer them. Instead, they look for candidates with an 'agentic mindset' who can use AI tools like Claude Code or Codex effectively. The company wants engineers who can build environments for AI agents to work better and verify their output. This shift focuses on proactive and curious individuals who prioritize solving user problems.

GitLab uses AI to fix code flaws and build pipelines

GitLab has released new AI features to improve software delivery and security. Agentic SAST Vulnerability Resolution can now automatically fix code flaws, preventing them from reaching production. New agents in the GitLab Duo Agent Platform can set up CI pipelines in minutes and provide quick answers from software lifecycle data. These tools help address bottlenecks in code delivery and operations. GitLab also introduced spending controls for AI usage to manage costs.

Canva uses AI to transform its design platform

Design platform Canva has launched Canva AI 2.0, a major update aimed at safeguarding its $42 billion valuation in the evolving AI landscape. The new tools allow its 265 million monthly users to generate visuals by describing what they want in conversational language. An AI agent will then build and iterate on the design. This rebuild positions Canva as a disruptor, rather than a target, in the face of AI advancements that threaten existing business models.

Globant expands AI role with Autodesk digital twins

Globant has expanded its collaboration with Autodesk, becoming an Autodesk Tandem Digital Twin Solution Provider. This partnership focuses on implementing digital twin technology for environments like airports and manufacturing facilities. Globant's expertise in AI and data analytics will be combined with Autodesk Tandem's platform to improve operational efficiency and decision-making. The move aims to help organizations manage physical assets more intelligently through advanced digital representations.

Sources

• OpenAI's Codex: Bridging AI and Code Development
• Codex for (almost) everything
• OpenAI Codex Gets Smarter, Broader
• Man used AI to make false statements in effort to shut down London nightclub
• Man used AI to make false statements to shut down London nightclub, police say
• Most of you are rejecting AI. The data shows you’re running out of time
• AI and Executive Protection: New Risks, New Defenses
• Palo Alto Networks founder applies to invest in California bank
• Erica Brescia: AI models need responsible development
• The CTO of an AI startup explains the shift in engineering hiring
• GitLab rolls out AI that fixes code flaws and builds pipelines
• Canva Leans Into AI to Defend Its $42 Billion Empire
• Is Globant’s (GLOB) Deeper Autodesk Tandem Role Quietly Reframing Its Physical AI Ambitions?