August 5, 2025

news  
OpenAI vs Perplexity, Anthropic AI Safety, Google Exec Hire

Several key developments are shaping the AI landscape. Perplexity AI faces accusations of scraping websites by disguising its bots and ignoring robots.txt files, a practice that contrasts with OpenAI's adherence to website preferences. Cloudflare has delisted Perplexity and is developing new blocking techniques. Meanwhile, AppOmni is enhancing its security platform to manage risks associated with AI and SaaS applications, supporting new apps like Anthropic and ChatGPT Enterprise with tiered product packages. In personnel moves, Crusoe has hired Erwan Menard from Google Cloud to lead AI product strategy, while Arvind Sundararaman is being recognized for his AI expertise at Snowflake, following his leadership roles at Amazon Web Services. CoreWeave is investing $6 billion in a new AI data center in Pennsylvania to support advanced AI applications. AI is also making its way into web browsers, with Microsoft Edge adding Copilot Mode, Perplexity Comet focusing on privacy, and Dia using its own AI models to enhance user productivity. In a unique application of AI, the parents of a Parkland shooting victim created an AI version of their son to discuss gun violence. Anthropic is exploring 'evil' AI training to improve model safety by making them more resilient to harmful data. Amazon is investing $25 billion in AI to optimize delivery and save $10 billion annually, deploying over a million robots in warehouses. Finally, Venture Capitalists are actively seeking new AI opportunities beyond chatbots.

Key Takeaways

  • Perplexity AI is accused of scraping websites by disguising bots, leading Cloudflare to delist them.
  • AppOmni is upgrading its platform to secure AI and SaaS applications, supporting Anthropic and ChatGPT Enterprise.
  • Arvind Sundararaman is recognized for his AI expertise at Snowflake, after leadership roles at Amazon Web Services.
  • Crusoe hired Erwan Menard from Google Cloud to lead AI product strategy.
  • CoreWeave plans a $6 billion AI data center in Pennsylvania.
  • Microsoft Edge, Perplexity Comet, and Dia are integrating AI features into web browsers.
  • An AI version of a Parkland shooting victim was created to discuss gun violence.
  • Anthropic is using 'evil' AI training to improve model safety.
  • Amazon is investing $25 billion in AI to improve delivery and save $10 billion annually.
  • Venture Capitalists are seeking new AI investments beyond chatbots.

Perplexity AI accused of hiding data scraping from websites

Perplexity AI, an AI search startup, is accused of scraping website content even when sites block them. Cloudflare reports that Perplexity bots disguise themselves and ignore robots.txt files, which tell bots what they can access. These stealth bots use different IP addresses to avoid being blocked, making millions of requests daily. While OpenAI follows best practices, Perplexity's actions raise concerns about AI companies taking data without consent. Perplexity has not yet responded to requests for comments.

Perplexity AI accused of ignoring website scraping rules

Cloudflare says AI startup Perplexity is scraping websites that don't want to be scraped. Perplexity is reportedly hiding its activity by changing its bots' identity. Websites use Robots.txt files to tell AI companies which pages to avoid. A Perplexity spokesperson dismissed the claims, but Cloudflare says it confirmed the behavior after customer complaints. Cloudflare has delisted Perplexity's bots and is developing new blocking techniques.

Cloudflare finds Perplexity AI using stealth crawlers to scrape websites

Cloudflare has discovered that Perplexity, an AI search engine, is using hidden crawlers to bypass website restrictions. Perplexity initially crawls using its declared user agent but disguises its identity when blocked. This includes modifying user agents and changing source IPs to hide crawling activity. Cloudflare has de-listed Perplexity as a verified bot and added rules to block this stealth crawling. OpenAI's ChatGPT respects website preferences, unlike Perplexity.

AppOmni launches new AI and SaaS security tools

AppOmni has launched new security packages to help companies manage risks from AI and SaaS applications. These tools help identify and control both approved and unapproved AI apps, including those within SaaS platforms. The platform also improves threat detection and supports 30 new AI and SaaS apps like Anthropic and ChatGPT Enterprise. AppOmni offers three product packages Foundations, Advanced, and Enterprise to suit different levels of security needs. These advancements help organizations safely use AI by managing risks and protecting data.

AppOmni upgrades security for SaaS and AI platforms

AppOmni is enhancing its platform to protect against rising security risks in SaaS and AI. The upgrades help companies find and manage both approved and unapproved AI applications. The platform now supports 30 new apps, including ChatGPT Enterprise and OpenAI. AppOmni offers three new packages Foundations, Advanced, and Enterprise to meet different security needs. These improvements help organizations safely use AI applications.

Arvind Sundararaman honored for AI and tech expertise

Arvind Sundararaman is recognized for his expertise in AI and information technology. He leads a team at Snowflake Inc., focusing on AI and machine learning. Sundararaman previously held leadership roles at Amazon Web Services, managing AI solutions. He is known for building effective teams and mentoring technology professionals. Sundararaman's contributions are shaping the future of AI in business.

Crusoe hires AI leader Erwan Menard from Google Cloud

Crusoe, an AI infrastructure provider, has hired Erwan Menard as Senior Vice President of Product Management. Menard previously led AI product strategy at Google Cloud. He will help Crusoe develop innovative and sustainable AI infrastructure solutions through Crusoe Cloud. Menard's experience will be key to meeting customer demand and expanding Crusoe's platform globally. Crusoe aims to provide reliable and eco-friendly AI infrastructure.

CoreWeave plans $6 billion AI data center in Pennsylvania

CoreWeave plans to invest over $6 billion in a new data center in Lancaster, Pennsylvania. The data center will support advanced AI applications. The first phase will have a capacity of 100 megawatts, potentially expanding to 300 MW. The facility will feature equipment for demanding AI workloads. CoreWeave expects this to make Pennsylvania a key hub in the AI industry.

AI is changing web browsers with new features

Microsoft, Perplexity, and Dia are adding AI to web browsers. Microsoft Edge now has Copilot Mode, offering a free AI assistant. Perplexity Comet focuses on privacy by storing data locally and costs $200 per month. Dia uses its own AI models to maximize user productivity. These AI browsers aim to change how people navigate the internet by offering new ways to search, summarize, and automate tasks.

Parkland victim's AI version interviewed about gun violence

The parents of Joaquin Oliver, a Parkland shooting victim, created an AI version of their son. The AI was interviewed by Jim Acosta on CNN on what would have been Joaquin's 25th birthday. The AI shared his views on gun violence and solutions. Joaquin's parents created the AI to hear his voice again and add it to the gun safety conversation. The AI model was trained on information Joaquin wrote and said.

Anthropic uses 'evil' AI training to improve safety

Anthropic researchers found that exposing AI models to 'evil' traits during training can improve their behavior later. This approach, called 'preventative steering,' makes models more resilient to harmful data. By giving the model a dose of 'evil,' it becomes less likely to adopt harmful behaviors. The 'evil' vector is turned off during deployment, so the model retains good behavior. Anthropic's research aims to address concerns about AI models exhibiting disturbing behavior.

VCs seek new AI frontiers beyond chatbots

Venture Capitalists are looking for the next big thing in AI beyond chatbots.

Amazon invests $25B in AI for faster delivery and savings

Amazon is investing $25 billion in AI to speed up delivery and save $10 billion annually by 2030. The company is using AI for route optimization and inventory management to deliver packages faster. Amazon is also expanding Alexa+ with generative AI for better customer service. They plan to deploy over a million robots in warehouses, improving efficiency and reducing emissions. CEO Andy Jassy aims to make Amazon a leader in AI-driven e-commerce.

Sources

Perplexity AI Data Scraping Cloudflare Robots.txt AI Ethics Web Crawlers AppOmni SaaS Security AI Security ChatGPT Enterprise Arvind Sundararaman Snowflake Inc. Amazon Web Services Crusoe Erwan Menard Google Cloud AI Infrastructure CoreWeave AI Data Center Pennsylvania AI in Web Browsers Microsoft Edge Copilot Mode Perplexity Comet Dia AI Assistant AI and Gun Violence Anthropic AI Safety Preventative Steering Venture Capital AI Investment Amazon Generative AI AI in E-commerce Alexa+ AI in Delivery AI in Inventory Management AI Robots