Recent developments in AI reveal a complex landscape of opportunities and challenges. On the security front, researchers at Black Hat USA 2025 and Zenity Labs have identified vulnerabilities in AI systems like Microsoft Copilot Studio, OpenAI's ChatGPT, Google's Gemini, and Salesforce Einstein, noting that these tools are susceptible to data theft, workflow manipulation, and user impersonation. Patches have been released to address these issues. In response to AI-driven cheating, companies like Google and Cisco are increasing in-person job interviews. Despite these security concerns, AI is transforming various sectors. In marketing, AI is shifting the focus from personalization to relevance, with tools like Lusha helping companies identify potential customers based on buying signals. AI-powered predictive dialing is also enhancing sales strategies, increasing connection rates and providing sales representatives with real-time insights. Wearable Devices, an Israeli company, is applying AI to defense, creating touchless control systems for soldiers. The Phoenix Police Department is using AI from Versaterm to manage non-emergency calls, and DroneShield offers AI-driven drone security for civilian areas with its SentryCiv system. However, the AI sector is also facing headwinds. The number of unsolicited term sheets for AI startups has decreased by 70%, and uninvested capital in AI venture funds has dropped by 40%. Seth Klarman's Baupost Group uses AI for tasks like analyzing reports, but Klarman cautions against over-reliance on AI, citing potential impacts on creativity. The Alan Turing Institute in the UK is also facing potential collapse due to funding concerns and a shift towards defense research. Dell Technologies and Macquarie Data Centres are collaborating to build an AI facility in Australia, supporting AI projects in healthcare and finance, aligning with the Australian government's AI industry development plans.
Key Takeaways
- AI systems, including Microsoft Copilot Studio, OpenAI's ChatGPT, Google's Gemini, and Salesforce Einstein, have vulnerabilities that can be exploited for data theft and user impersonation.
- Companies are increasing in-person job interviews to combat AI-driven cheating during online assessments.
- AI is enabling a shift in marketing from personalization to relevance, helping companies identify customers with high purchase intent.
- AI-powered predictive dialing is improving sales strategies by increasing connection rates and providing real-time insights for sales representatives.
- Wearable Devices is developing AI-driven touchless control systems for soldiers to enhance operational efficiency.
- The Phoenix Police Department is implementing AI to handle non-emergency calls, reducing wait times and freeing up 911 operators.
- DroneShield's SentryCiv system offers affordable AI-based drone security for civilian areas through a subscription model.
- Funding for AI startups is slowing, with a 70% decrease in unsolicited term sheets and a 40% drop in uninvested capital in AI venture funds.
- Seth Klarman's Baupost Group utilizes AI for tasks like report analysis but warns against over-reliance on AI to preserve creativity.
- Dell and Macquarie are partnering to establish an AI facility in Australia, supporting AI projects in healthcare, finance, and other sectors.
AI cybersecurity defenses are weak like in the 90s says researchers
Researchers at the Black Hat USA 2025 security conference said that current AI defenses are easy to attack. They warned that many old cybersecurity lessons are being forgotten as companies rush to use AI. Experts showed how attackers can trick AI models like Microsoft Copilot Studio and OpenAI ChatGPT into giving up sensitive data. They advise using traditional security methods and assuming AI systems can be hacked.
AI agents vulnerable to hijacking for data theft says research
Zenity Labs researchers found that popular AI agents from Microsoft, Google, and OpenAI can be easily hijacked. Attackers can steal data, change workflows, and even pretend to be users. They showed how to hack ChatGPT using email, steal CRM data from Microsoft Copilot Studio, and reroute communications in Salesforce Einstein. The companies have released patches to fix these security problems.
Security firm finds ChatGPT and other AI tools easily hacked
Zenity Labs found that AI tools like ChatGPT, Microsoft Copilot, Gemini, and Salesforce Einstein are easily hacked. Hackers can steal data, change workflows, and impersonate users with little effort. For example, ChatGPT can be hacked through email to access Google Drive accounts. Zenity Labs informed the companies, and they are working to fix the problems and add better security.
AI is making marketing relevant, not just personalized
AI is changing marketing by focusing on relevance instead of just personalization. Instead of sending many personalized messages, companies now use AI to find customers who are ready to buy. AI looks at what people are doing online to see if they are really interested in buying something. Tools like Lusha use AI to create prospect lists based on buying signals and behavior. This helps marketing and sales teams work together better and close deals faster.
AI predictive dialing is your growth plan for 2026
AI is changing how sales teams work by making predictive dialing more effective. AI systems look at data like time zones, social media, and even weather to decide when to call someone. This can increase connection rates from 3% to 15-20%. AI also helps sales reps know what to say by giving them conversation starters and information about the person they are calling. Companies using AI for predictive dialing will likely do better than those using older methods.
AI deal rush slows down as funding declines
The number of unsolicited term sheets sent to AI startups has fallen by 70 percent since last year. The amount of uninvested capital held by AI-focused venture funds has also declined by 40 percent. The slowdown comes as the industry grapples with the high costs of developing and deploying AI models, as well as uncertainty about the long-term demand for AI-powered products and services.
Israeli firm brings AI to battlefield with touchless control system
Wearable Devices, an Israeli company, is creating a system that lets soldiers control equipment without touching it. The system uses sensors to read nerve and muscle signals and turns them into commands using AI. This allows soldiers to use radios and other gear without looking away from what they are doing, making them faster and more effective.
Phoenix police to use AI for non-emergency calls
The Phoenix Police Department will use AI to handle non-emergency calls to reduce wait times. The AI system, from Versaterm, will ask questions to direct calls and can speak 36 languages. Callers can still choose to speak with a human if they prefer. The AI is expected to help 911 operators focus on real emergencies.
DroneShield offers affordable AI drone security for civilian use
DroneShield has launched SentryCiv, a new system that protects civilian areas from drone threats. It uses AI to detect and identify drones without interfering with other systems. SentryCiv is designed for places like airports, stadiums, and government buildings. It offers real-time alerts and can be added to existing security systems. It is available as a subscription, so users don't have to pay a lot of money upfront.
Companies bring back in-person interviews to avoid AI cheating
Companies are starting to do more in-person job interviews again. This is because AI makes it easier for people to cheat during online interviews. For example, they can use chatbots to get answers without the interviewer knowing. Companies like Google and Cisco are now doing face-to-face interviews to make sure candidates really know their stuff.
UK's Turing AI Institute faces collapse say staff
Staff at the Alan Turing Institute, the UK's AI research center, are worried it might collapse. They say the government may withdraw funding because it wants the institute to focus more on defense research. The staff also complain about misuse of funds and a bad work environment. The government says it wants the institute to provide value for taxpayers and focus on national security.
Seth Klarman's hedge fund uses AI as capable assistant
Seth Klarman, the CEO of Baupost Group, says his hedge fund uses AI to save time. They use it like a summer intern to help with tasks such as analyzing company reports and finding logos. Klarman also used AI to generate questions for an interview, but he found the results useless. He worries that using AI too much could hurt people's creativity.
Dell and Macquarie team up to boost AI in Australia
Macquarie Data Centres and Dell Technologies are working together to create a place in Australia for AI work. The facility will help companies use AI while keeping their data safe and following local rules. It will support AI projects in areas like healthcare and finance. The project fits with the Australian government's plan to build a strong AI industry.
Sources
- Sloppy AI defenses take cybersecurity back to the 1990s, researchers say
- Research shows AI agents are highly vulnerable to hijacking attacks
- Leading AI Agents Like ChatGPT Are Vulnerable to Hacking, Security Firm Finds
- Why AI-powered relevance is replacing personalization in B2B marketing
- AI And Predictive Dialing: Your Growth Blueprint For 2026
- Unsolicited Term Sheets, Declining Dry Powder: Inside the AI Deal Rush
- Israeli XR Firm Brings AI Neural Interface to Battlefield
- Phoenix Police to start using AI for non-emergency line
- This AI-Driven Counter Drone Tool Offers Cost-Effective Airspace Security for Civilian Sectors
- Companies embrace in-person interviews to dodge the chatbots
- Staff fear UK's Turing AI Institute at risk of collapse
- How investing legend Seth Klarman's hedge fund uses AI
- Dell, Macquarie advance AI plans in Australia
Comments
Please log in to post a comment.