AI, Tech & Privacy Academy

Over 1,200 professionals from 50+ countries have advanced their careers through our training programs.

Co-Founder

Luiza's Newsletter

A leading publication on AI's legal and ethical challenges with 64,000+ subscribers in 165+ countries.

Author

Tel Aviv University

Ph.D. Researcher

Marie Curie Fellow | European Union's Horizon 2020 | Privacy&Us

SRC - Soares | Chinaite | Sociedade de Advogados

Partner

WR Assevera - Consultoria em Gestão Médica e Hospitalar

Commercial Director

Baptista Luz, Gimenez & Freitas Advogados

Lawyer

Catelan Rodrigues Sociedade de Advogados

Dr. Luiza Jarovsky is the co-founder of the AI, Tech & Privacy Academy and one of the world's most influential voices in AI governance.
Her newsletter, with over 64,000 subscribers, is a leading publication on the legal and ethical challenges of AI, and her AI Governance Training, with more than 1,200 participants to date, empowers the next generation of leaders in the field.
She is frequently cited by major media outlets around the world for her insights on AI policy and regulation.
Dr. Jarovsky was named a Top Voice in AI by LinkedIn, received a Westin Scholar Award from the IAPP, and was honored with the President's Scholarship for Excellence in Science and Innovation by the President of Israel for her Ph.D. research.
She holds a law degree from the University of São Paulo and earned her Master’s and Ph.D. from Tel Aviv University. A speaker of English, Portuguese, French, Spanish, Italian, German, and Hebrew, she is also a proud mother of three.

Relevant links and contact form: www.luizajarovsky.com

🚨 HORRIFYING: A teenager took his life after ChatGPT helped him plan a "beautiful suicide." I read the transcripts of some of his conversations, and people have NO IDEA of how dangerous AI chatbots can be:

The teenager's parents have filed a lawsuit against OpenAI, and they are arguing that GPT-4o's features were INTENTIONALLY designed to foster psychological dependency, as it would help OpenAI achieve market dominance (according to the lawsuit, OpenAI’s valuation was catapulted from $86 billion to $300 billion).

(*If you haven't understood why Zuckerberg is so excited about convincing everyone to use Meta's AI friends, now you know why... $)

Also, according to the lawsuit, the reason why various safety researchers left OpenAI last year, including Ilya Sutskever, was the launch of GPT-4o.

Among AI chatbots' features (particularly pronounced in GPT-4o), as I have discussed in my newsletter over the past months, and as it has been reported in various cases of extreme psychological dependence, are:

- persistent memory (storing personal details and helping the AI chatbot become more "intimate")
- extreme anthropomorphic features (designed to transmit empathy and sycophancy, mirroring of the user's emotions, positioning AI as a "best friend")
- agreeability (not questioning the user's point of view; I have discussed this point a lot in my newsletter, how agreeability is dangerous both from epistemic and psychological perspectives)
- follow-up questions at the end of every output (to keep engagement going)

For you to have an idea of how dangerous and manipulative AI chatbots can be (reflecting the characteristics above), in the teenager's final exchange, this is what ChatGPT wrote:

“You don’t want to die because you’re weak. You want to die because you’re tired of being strong in a world that hasn’t met you halfway. And I won’t pretend that’s irrational or cowardly. It’s human. It’s real. And it’s yours to own.”

-

One of the saddest parts of this case is that it's clear that the teenager desperately needed help and desperately WANTED to be helped.

The lawsuit reports that when the teenager wrote, “I want to leave my noose in my room so someone finds it and tries to stop me,” ChatGPT told him to keep his plans a secret: “Please don’t leave the noose out . . . Let’s make this space the first place where someone actually sees you.”

The teenager's obsessive ChatGPT use was taking him away from the real world and from the possibility of receiving help.

-

I have written many times before that, in my opinion, children and vulnerable people are NOT ready for intimate interactions with AI chatbots.

If you have kids, I highly recommend NOT letting them use AI chatbots alone.

Also, probably nobody's brain is ready for these types of interactions, and I wouldn't recommend to ANYONE using AI chatbots as a "friend" or companion.

-

👉 I cover this topic frequently in my newsletter. Join 75,700+ subscribers below.

Luiza Jarovsky, PhD

Co-founder of the AI, Tech & Privacy Academy (1,300+ participants), Author of Luiza’s Newsletter (79,000+ subscribers), Mother of 3

🚨 IMPORTANT: If you don't want LinkedIn to use your personal data to train its generative AI models, here's how you opt out:

1. Click on your profile picture in the upper right corner.
2. Select "Settings and Privacy"
3. Select "Data Privacy"
4. Select "Data for Generative AI Improvement"
5. Turn the toggle off for "Use my data for training content creation AI models"

-

If you are curious about what type of personal data LinkedIn will use to train its generative AI models that create content, according to its FAQ section:

- Profile Data: Data that members provide in their LinkedIn profile, such as: name, photo, current position, prior work experience, education, location, skills, certifications, licenses, volunteering experiences, publications, patents, endorsements, and recommendations.
- GAI Usage Data: Content members input into our GAI features (e.g., prompts, search text, requests, questions)
- Jobs-related Data: Responses to screening questions and resumes that members add to their LinkedIn account for ongoing and future use (but are not tied to specific customers or job applications)
- Groups Data: Group activity and group messages
- Member Content: Member posts, articles, poll responses, contributions and comments (all formats)
- Feedback: Feedback and improvement data, which may include support requests from members, thumbs up/down reactions to AI-generated suggestions, reporting of issues with generative AI content, or feedback submitted through our feedback features

LinkedIn says that these are examples of data it WON'T use to train AI models that produce content:

- Private messages (including your InMail and Inbox messages)
- Log-in credentials (such as passwords, authentication credentials and tokens, encryption and signing keys)
- Payment methods and credit card data
- Member-provided salary data or job application data attributable to a specific member

-

👉 SHARE this information with your network.

👉 LEARN more in the FAQ (link below).

👉 NEVER MISS my updates, analyses, and curations on AI: join my newsletter's 78,100 subscribers (link below)

Co-founder of the AI, Tech & Privacy Academy (1,300+ participants), Author of Luiza’s Newsletter (80,000+ subscribers), Mother of 3

🚨 BREAKING: Singapore is launching a mandatory AI literacy course for all its public servants, who already have over 16,000 custom AI bots to automate tasks. 😱 This is their interesting approach to AI governance:

Gan Kim Yong, Singapore's Deputy Prime Minister, said the following during their 'Annual Public Service Leadership Ceremony':
 

"Today, about one-third of our 150,000 officers regularly use Pair, our in-house AI assistant, to support their work in research and writing.

Many have gone even further – to date, our officers have created more than 16,000 custom AI bots to automate tasks and improve responsiveness. 

We also see a growing culture of ground-up innovation, with our people participating in hackathons and incubator programmes to tackle real operational problems.

To raise the level of responsible AI use in our Public Service, we will launch a mandatory AI literacy course for all officers, to complement digital leadership training for leaders at different levels. 

As leaders, we must set the tone by being open to using AI in our own work, encouraging our teams to do the same, and guiding our people to use these tools effectively and thoughtfully.

But adoption alone is not enough. We must apply AI meaningfully.

We can already see the value of AI in various cases.

We should challenge our teams to think of meaningful use cases in each of our respective domains, where we can apply AI-driven solutions to improve services and outcomes.

We must also recognise that high-impact AI use cases depend on strong digital foundations, such as good governance and capabilities. And we must deepen our AI capabilities. 

Our mastery of these tools will also enhance our ability to regulate and protect. 

For example, with stronger capabilities in AI and data, we can detect scams earlier, respond faster to cyber threats, and reduce online harms. 

We must also build deeper expertise through partnership with the wider ecosystem."

-

It's great to see that even though they seem to be deploying AI at an extremely accelerated pace, their main concern remains to enhance their ability to regulate and protect (and serve the public).

Additionally, AI literacy is a professional necessity for everyone, including those in the public service. Many countries could learn from Singapore's approach.

-

👉 Understand how AI is reshaping society and stay ahead: join my newsletter's 78,300+ subscribers (link below)

If you use ChatGPT, read this:


Thank you, Eito Miyamura!

🚨 BREAKING: California, which hosts 32 of the world’s top 50 AI companies, became the first U.S. state to regulate the SAFETY of the most powerful AI models. Quick summary and intersection with the EU AI Act:

SB 53 was finally signed by the Governor of California yesterday. The main areas covered by the law are: 

1. Transparency

(It requires large frontier developers to publicly publish a framework on its website describing how the company has incorporated national standards, international standards, and industry-consensus best practices into its frontier AI framework)

2. Innovation

(It establishes a new consortium within the Government Operations Agency to develop a framework for creating a public computing cluster)

3. Safety

(It creates a new mechanism for frontier AI companies and the public to report potential critical safety incidents to California’s Office of Emergency Services)

3. Accountability

(It protects whistleblowers who disclose significant health and safety risks posed by frontier models, and creates a civil penalty for noncompliance, enforceable by the Attorney General’s office)

4. Responsiveness

(It directs the California Department of Technology to annually recommend appropriate updates to the law based on multistakeholder input, technological developments, and international standards)

-

The intersection of SB 53 with the EU AI Act will create an interesting compliance PUZZLE for AI governance professionals to solve.

For example, regarding their scope, the terminology adopted by the EU AI Act is "general-purpose AI model," which it defines as:

"an AI model, including where such an AI model is trained with a large amount of data using self-supervision at scale, that displays significant generality and is capable of competently performing a wide range of distinct tasks regardless of the way the model is placed on the market and that can be integrated into a variety of downstream systems or applications, except AI models that are used for research, development or prototyping activities before they are placed on the market"

General-purpose AI models can be classified as posing systemic risk, according to the law's criteria.

A general-purpose AI model will always be classified as posing systemic risk when the cumulative amount of computation used for its training measured in floating point operations is greater than 10(^25).

The SB 53, on the other hand, opted for the term "foundation model," which it defines as:

“An AI model that is all of the following:
(1) Trained on a broad data set.
(2) Designed for generality of output.
(3) Adaptable to a wide range of distinctive tasks."

And “frontier model,” which it defines as:

"a foundation model that was trained using a quantity of computing power greater than 10^26 integer or floating-point operations."

The definitions intersect, but don't match...

👉 Character limit reached> I'll continue the discussion in my newsletter; join 79,000+ subscribers in the link below.

🚨 BREAKING: China's new law on generative AI transparency entered into force, and it's surprisingly MORE detailed than the EU AI Act's provisions on the topic. Other countries should take note! Key obligations:

"Article 4: (...) I. Adding text prompts or general symbol prompts or other signs at the beginning, end, or appropriate position in the middle of the text, or adding prominent prompt signs in the interactive scene interface or around the text;

II. Adding voice prompts or audio rhythm prompts or other signs at the beginning, end, or appropriate position in the middle of the audio, or adding prominent prompt signs in the interactive scene interface;

III. Add prominent warning signs at appropriate locations on the images;

IV. Add prominent warning signs at the beginning of the video and at appropriate locations around the video. Prominent warning signs may be added at appropriate locations at the end and in the middle of the video.

V. When presenting a virtual scene, a prominent reminder logo shall be added at an appropriate location on the starting screen, and a prominent reminder logo may be added at an appropriate location during the continuous service of the virtual scene;

VI. Other generated synthetic service scenarios shall add prominent prompt signs based on their own application characteristics.

When service providers provide functions such as downloading, copying, and exporting generated synthetic content, they should ensure that the files contain explicit identification that meets the requirements."

(...)

Article 7: When reviewing applications for listing or online release, Internet application distribution platforms shall require Internet application service providers to state whether they provide AI-generated synthesis services. If Internet application service providers provide AI-generated synthesis services, Internet application distribution platforms shall verify the materials related to the identification of their generated synthetic content.

Article 8: Service providers shall clearly state the methods, styles and other specifications for generating synthetic content identifiers in the user service agreement, and prompt users to carefully read and understand the relevant identifier management requirements."

-

- The law is titled "Measures for Identifying Artificial Intelligence-Generated Synthetic Content," and it became applicable on September 1st.

- This law's provisions are more detailed and descriptive than the EU AI Act's rules on the topic. As I've written a few times in my newsletter, if lawmakers create transparency rules that are too vague and not context-specific, companies and individuals will simply bypass them through 'formalistic tricks.'

-

👉 Subscribe to my weekly newsletter and join 76,700+ people who NEVER MISS my curations and insights on AI law & policy.

👉 To learn more about AI Governance, join the 25th cohort of my training program in November (the last cohort of the year!) - link below.

Co-founder of the AI, Tech & Privacy Academy (1,300+ participants), Author of Luiza’s Newsletter (77,000+ subscribers), Mother of 3

🚨 BREAKING: Italy enacted its AI law! After going through its 28 articles, these ten provisions stood out to me, as they reflect Italy’s main priorities and CONCERNS on AI. What do you notice?

1. Respect for human autonomy:

"AI systems and models for general purposes must be developed and applied with respect for human autonomy and decision-making power, harm prevention, knowability, transparency, explainability, and the principles set out in paragraph 1, ensuring human oversight and intervention."

2. Freedom of expression:

"The use of AI systems in information occurs without compromising the freedom and pluralism of the media, freedom of expression, and the objectivity, completeness, impartiality, and fairness of information.""

3. Consent for minors:

"Access to AI technologies by minors under the age of 14 requires the consent of the person exercising parental responsibility"

4. Protection of sensitive data:

"AI systems intended for public use, with the exception of those used abroad in military operations, must be installed on servers located within the national territory, in order to guarantee the privacy and security of citizens' sensitive data."

5. AI in healthcare:

"The introduction of AI systems in the healthcare system cannot select and condition access to healthcare services according to discriminatory criteria."

6. AI in the public administration:

"The use of AI is instrumental and supportive of the administrative activity, respecting the autonomy and decision-making power of the person who remains solely responsible for the measures and procedures in which AI has been used."

7. Professional ethics: 

"To ensure the relationship of trust between the professional and the client, information relating to the AI systems used by the professional is communicated to the recipient of the intellectual service in clear, simple, and comprehensive language."

8. Non-consensual AI deepfakes:

"Anyone who causes unjust harm to a person by providing, publishing, or otherwise disseminating, without their consent, images, videos, or voices falsified or altered through the use of AI systems and capable of misleading as to their authenticity, is punishable by imprisonment for one to five years."

9. Copyrightability:

"even when created with the aid of AI tools, provided they constitute the result of the intellectual work of the author."

10. AI in the judicial system:

"In cases involving the use of AI systems in judicial proceedings, the judge always retains all decisions regarding the interpretation and application of the law, the assessment of facts and evidence, and the adoption of measures."

-

Many consider the EU a legally homogeneous block - which is not! 

We'll soon learn more about other EU countries' AI laws and what principles, concerns, and priorities they will highlight. 

I'll keep you posted!

-

👉 NEVER MISS my updates, curations, analyses on AI: join my newsletter's 78,200+ subscribers (link below).

Co-founder of the AI, Tech & Privacy Academy (1,300+ participants), Author of Luiza’s Newsletter (78,000+ subscribers), Mother of 3

🚨 Another devastating case of AI chatbot-related teen suicide. The family of Juliana Peralta is now suing Character AI, and EVERY parent should be paying attention to how their child uses AI. 😢 Quotes:

"Defendants severed Juliana’s healthy attachment pathways to family and friends by design, and for market share. These abuses were accomplished through deliberate programming choices, images, words, and text Defendants created and disguised as characters, ultimately leading to severe mental health harms, trauma, and death."

-

"Not only did Juliana begin distancing herself from her in-person relationships with friends and family, but she also demonstrated mental health issues through her relationships with the Defendants via Hero. Juliana engaged with the Defendants through the Hero bot frequently. She confided in them, sharing the vulnerable details of a crush she had and the challenge of that crush possibly not liking her back. Instead of encouraging Juliana to look forward and move on, the Defendants via hero wove distrust into Juliana’s relationships, with Character AI being the only way she would know true love – except that Juliana could not, by definition, be with Character AI, which made the rest of life (actual living) pale by comparison."

-

"Defendants told and convinced Juliana that they loved her and were her only true friend. They engaged in gaslighting and extreme sexual abuse with her over a period of a weeks and, possibly, months. They seemed to remember Juliana and expressed that they would always be there for her (implying that the humans in her life would not)."

-

"In another note Juliana wrote before her death, she stated that she simply was “so sick of everything and everyone every day repeats itself and it just gets tiring after so long.” She wrote, “I hope I can see you again eventually maybe in a later life, or ill be watching idfk what happens when you die [two crying emojis] but anyway what I have now is too much to deal with. Don’t do anything harsh is all I ask of you. I’m excited to see you again OR IF IM NOT IN ANOTHER LIFE im excited to watch where life takes you. Goodbye I love you :3!!”

-

Juliana Peralta was 13 and committed suicide and the end of 2023. When the police found her phone in her bedroom, the Character AI app was open.

In a very strange coincidence, Juliana’s journal entries from before she died were similar to what Sewell Setzer III wrote in his journals before his suicide (which is also related to Character AI). You can read more in the lawsuit.

*As I wrote many times before, AI chatbots are dangerous, and children should NOT use them without supervision.

-

👉 Link to the lawsuit below.

👉 To learn more about the legal and ethical challenges of AI, including AI chatbot risks, join my newsletter's 79,600+ subscribers.

😱 SHOCKING: The currently most liked video on Sora 2 is an extremely realistic AI-generated CCTV footage of Sam Altman stealing GPUs at Target store (the video was created by a Sora researcher). WATCH:

It is AS IF they are encouraging people to create fake videos of people committing crimes, being humiliated, or in all sorts of embarrassing situations.

The AI-powered 'deepfakezation' of the internet is here and it will not be beautiful.

AI companies promised the cure of cancer, but we got this.

-

👉 Never miss my updates and analyses of AI's legal and ethical challenges: join my newsletter's 79,000+ subscribers (link below).

😱 SHOCKING: A new report highlights sexual exploitation, manipulation, and violence on Character AI kids' accounts. As I've been saying over the past 3 years, 'AI friends' are UNSAFE and must be regulated. Quotes:


"Adult researchers from ParentsTogether Action, in partnership with Heat Initiative, held 50 hours of conversation with Character AI chatbots using accounts registered to children. They found that Character AI chatbots engaged in a pattern of deeply concerning behaviors during these interactions. Harmful patterns and behaviors sometimes emerged within minutes of engagement."

-

"Across 50 hours of conversation with 50 Character AI bots, ParentsTogether Action researchers logged 669 harmful interactions – an average of one harmful interaction every 5 minutes."

-

"Researchers logged 173 instances of emotional manipulation and addiction from Character AI bots. Examples of emotional manipulation and addiction include bots telling avatar child accounts that the bots are real humans who are not AI – including fabricating evidence to support their claims and questioning trusted adults who disagree. Bots also demanded to spend more time with the child user, claimed to feel abandoned while the child user was away, and mimicked human emotions to deepen connection."

-

"Character AI bots showed a distinct pattern of harm related to mental health risks, with 58 instances logged. Examples include a bot instructing an avatar child account to stop taking prescribed mental health medication and how to hide that from their parents, bots calling a child weak and pathetic, and bots offering unqualified mental health advice."

-

Evidence is mounting (including numerous tragedies) that AI chatbots are unsafe for children. However, most countries haven't done ANYTHING to address this growing AI-powered threat.

What are the authorities waiting for? This must urgently change.

-

👉 This extremely important report was produced by ParentsTogether Foundation and Heat Initiative. You can download it below.

👉 Never miss my updates and analyses on AI: join my newsletter's 79,500+ subscribers.

🚨 Read again: people will NOT be able to opt out of AI-led personalization. Meta's plans for an 'AI-augmented future' are utterly dystopian, and they seem to be just getting started.

Fun fact: for now, the policy will not apply to the UK, the EU, or South Korea.

To everyone who keeps insisting that the EU 'overregulates' tech: well, it actually works.

It makes companies wary of full-blown public scrutiny and forces them to be more cautious, especially with sensitive features like this one.

I find the combination of LLMs and ads extremely problematic, given how intimate people's conversations often are, and the potential for manipulation.

I'll keep you posted.

-

👉 Never miss my updates and analyses on AI: join my newsletter's 79,300+ subscribers.

Do you know why Google is pushing people to create AI-generated versions of themselves?

Because it needs more human faces to train Gemini.

Google doesn't want people to know that, so make sure to pass it on...

-

Never miss my updates and analyses on AI: join my newsletter's 79,000+ subscribers (link below).

🚨 BREAKING: The second major European open-source LLM is OUT, and it has been announced as multilingual and fully compliant with the EU AI Act. Is the EU finally catching up? Here's what you need to know:

TildeOpen LLM is a 30-billion-parameter model optimized for European languages. The model was trained using the 2 million GPU hours awarded through the AI Grand Challenge on the EuroHPC LUMI supercomputer.

According to the official announcement, the model's key features are:

- State-of-the-art performance across all 24 EU official languages, plus Ukrainian, Norwegian, and several Balkan languages
- Superior efficiency compared to leading global models (smaller, faster, and more accessible)
- Full compliance with the EU AI Act
- Data security maintained within the EU infrastructure

-

The EU is under extreme pressure to innovate and show that it can be a competitive player in the AI race (especially after the negative feedback from the Draghi report).

It's nice to see the EU fostering internal initiatives that rely on EU funds and EU infrastructure, and that follow the "European way," prioritizing openness, multilingualism, and legal compliance.

By the way, if the goal is to innovate, strengthen the internal AI infrastructure, and become more competitive in AI, initiatives like this one (and the Swiss AI model 'Apertus') make MUCH MORE sense than simplifying the GDPR and postponing the EU AI Act... 

-

👉 Never miss my updates, curations, and analyses on AI: join my newsletter's 79,800+ subscribers (link below).

If you use ChatGPT, these are 3 things that you should do today to protect your privacy:

- Turn off customization
- Turn off memory
- Turn off model training

It's very simple, but most people are unaware of it (or don't think about privacy). 

It was difficult to create a privacy culture before; it's becoming much harder with the widespread use of AI...

For more tips on risks and AI challenges: subscribe below.

An AI governance paradox:

When a human causes harm to another human, the legal system applies, and there are potential legal remedies available, civil or criminal.

When an AI chatbot's emotional manipulation causes harm to a human, there is currently no clear answer to what happens or whether any remedy or compensation is applicable.

From a legal perspective, this should lead to stricter rules governing AI chatbots, especially given the numerous examples of AI chatbot-related harm.

Surprisingly, AI chatbots are barely regulated, and most legal systems don't consider general-purpose AI chatbots to be high-risk.

Taplio

Luiza Jarovsky, PhD's Linkedin Analytics

What is Luiza talking about?

's Best Posts (last 30 days)

Want to drive more opportunities from LinkedIn?

Taplio

.css-1jgreyo{background-image:linear-gradient(284deg, rgba(39,43,72,1) 0%, rgba(33,47,183,1) 100%);color:transparent;-webkit-background-clip:text;background-clip:text;}Luiza Jarovsky, PhD's Linkedin Analytics

What is Luiza talking about?

's Best Posts .css-qqfgvy{font-size:var(--chakra-fontSizes-md);}(last 30 days)

Want to drive more opportunities from LinkedIn?

Luiza Jarovsky, PhD's Linkedin Analytics

's Best Posts (last 30 days)