Confident Security Emerges with $4.2M to Secure AI Data ...

For privacy and security, think twice before granting AI ...

McDonald's AI hiring chatbot exposed data of job candidates

AI in Cybersecurity: How AI is Changing Threat Defense

Navigating confidentiality risks in third-party AI tools

Top Data Privacy & AI Developments of 2025: Midyear ...

AI-powered attacks rise as CISOs prioritize AI security risks

AI and Tech under the One Big Beautiful Bill Act

Security, AI Oversight Are Flashpoints in Draft Defense Bill

A new company called Confident Security came out of hiding with $4.2 million to build special locks for AI data [1]. Their technology uses **end-to-end encryption** to stop AI companies from saving your questions or personal details. This helps companies follow strict European Union rules called the **AI Act** that protect people's safety and privacy [1]. In the United States, President Trump signed the **One Big Beautiful Bill Act** on July 4, 2025 [8]. This new law gives money to American AI companies but has strong rules about working with foreign countries, especially China. It also requires AI companies to use more materials made in the U.S. [8].

Security experts warned everyone to think twice before letting AI helpers access personal information [2]. When you let an AI book a restaurant table, it might ask for your calendar, contacts, and credit card – like giving away a photo of your private life. Signal app president Meredith Whittaker said using these AI helpers is like "putting your brain in a jar" because you lose control of your secrets [2]. Perplexity AI says it stores data on your device, but you still let them use your information to improve their models [2].

McDonald's had a security problem with its AI hiring chatbot called McHire [3]. Security researchers found weak passwords in the system that let them see five U.S. job applicants' private details like names and phone numbers [3]. The chatbot was made by a company called Paradox.ai, which fixed the problem within hours. McDonald's said they were disappointed and made the company solve the issue immediately [3]. This shows why companies must keep AI tools updated with strong passwords.

AI is becoming a double-edged sword for cybersecurity. On the good side, **autonomous response systems** can stop cyber attacks instantly without waiting for humans [4]. For example, they can block ransomware attacks right away, saving companies an average of **$2.22 million** [4]. Another helpful tool is **federated learning**, where AI learns from data without moving private information from its original location [4]. But bad news: 1 in 4 security bosses reported AI-powered attacks on their companies this year [7]. These attacks are hard to spot because they act like humans. Security leaders now worry more about AI risks than other dangers [7].

Companies are being told to create clear rules for using AI tools safely [5]. About 25% of American workers use AI weekly, with over 10% using it daily for tasks like writing or data analysis [5]. Lawyer Andrew Tibbetts says every company needs policies about what information can be shared with which AI tools to avoid leaking secrets [5]. Almost 70% of companies already use AI helpers, with 23% planning to start next year [7]. Security bosses want simple "allow-by-default" controls to safely manage these tools [7].

New protections are being developed for future threats. AI is helping create **quantum-resistant cryptography** to prepare for super-powerful quantum computers that could break today's digital locks [4]. In California, a big settlement against Healthline Media showed that states are strictly enforcing privacy rules, especially for medical information [6]. Lawyer Heather Egan called this a "wake-up call" for all websites to protect consumer data properly [6].

This week brought important news about keeping data safe with AI. Confident Security got $4.2 million to build special locks for AI data so companies can't see your private questions or information [1]. Experts warned people to be careful about letting AI helpers see personal stuff like calendars and messages because it's like giving away your brain in a jar [2]. McDonald's had a problem where its AI hiring chatbot showed some job applicants' private details because of old passwords, but it was fixed quickly [3].

New U.S. rules called the One Big Beautiful Bill Act now control how AI companies work with foreign countries and require more American-made parts [8]. Almost 25% of security bosses said bad people used AI to attack their companies this year, making AI safety their top worry [7]. Companies are being told to set clear rules about what workers can share with AI tools to protect secrets [5].

Good news: AI is also helping stop cyber attacks faster than humans can. New systems can block dangers automatically, saving companies money [4]. Special learning methods let AI get smarter without moving private data around, keeping information safer [4].

Data Privacy & Security Weekly AI News

Specific Topics