.svg)
.svg)
.svg)
.svg)
.svg)
.svg)
.svg)
.svg)
.svg)
.svg)
.svg)
.svg)
.svg)
.svg)
.svg)
.svg)
.svg)
.svg)
.svg)
.svg)
Dean Levitt
Flaw in Google Gemini allows email summaries to be used for phishing
Farah Amod : July 23, 2025
Researchers have shown that attackers can manipulate Gemini-generated email summaries to deliver hidden phishing prompts.
Hello world!
Today’s Paubox Weekly is 396 words - a 2 minute read.
Want to get this type of content delivered to your inbox every Friday? Subscribe to Paubox Weekly.
1. How to spot AI phishing attempts and other security threats
The rise of artificial intelligence has transformed our daily lives already and cyber threats are no exception. Cybercriminals have started leveraging AI to generate more sophisticated phishing emails and other email security threats that are increasingly difficult to detect.
In the know: AI-generated phishing emails have several distinguishing features that differentiate them from traditional phishing attempts.
7 ways to detect an AI phishing attempt
2. Sam Altman, GPT-5 and the future of AI
Speaking at MIT, Altman discussed GPT-5's development, the changing importance of large language model size, and OpenAI's current focus on expanding GPT-4's capabilities.
Why it matters: Efficient AI models could lead to faster analysis of medical data, enabling healthcare professionals to make quicker decisions in time-sensitive situations, such as emergency care or surgery.
Training matters more than larger data sets
3. Report shows increasing ransomware and lawsuits for pixel use
BakerHostetler released its annual Data Security Incident Response Report on the latest trends in cybersecurity. 24% of cases handled by the firm are in the healthcare industry.
Worth noting: Ransomware resulted in delayed care for patients with devastating consequences. The report found it took an average of 10.3 days for an organization to operate at an "acceptable" capacity again.
The average ransom paid is now $1,562,141
4. Amazon’s new clinic may create privacy loopholes
The new, low-price health service requires authorization from patients that may allow for sensitive data to be used or spread by Amazon.
Between the lines: The Washington Post reported that to receive services users must agree to the "use and disclosure of protected health information." Amazon states that information may be "re-disclosed," which may no longer be protected by HIPAA.
Could PHI be used for a targeting?
Community links
- Major changes expected in HIPAA law this spring. Link
- How to implement a HIPAA compliant privacy policy in your practice. Link
- Does HIPAA apply to employers? Link
- What is the HIPAA need to know rule? Link
- Is it a HIPAA violation to email patient names? Link
Good reads from around the web
- Generative AI could augment 40% of healthcare working hours. Link
- 3 things you should know about the end of the Covid public health emergency. Link
- What the HIPAA? Demystifying healthcare regulations for startups. Link
- Could wearables capture well-being? Link
- The pizza paradox: Unraveling the complexity of digital innovation in healthcare. Link
Top Videos
New AI attack uses hidden prompts in images to steal user data
Farah Amod : September 2, 2025
A new exploit can trick AI systems into executing hidden instructions by embedding them in images processed before reaching the model.
AI-powered ‘SpamGPT’ toolkit supercharges phishing attacks at scale
Farah Amod : October 2, 2025
A dark web tool mimicking professional marketing software is giving cybercriminals new powers to launch large-scale, AI-generated phishing campaigns.