Morrissey Technology

Loading

Tag ChatGPT

OpenAI
News

OpenAI Was Hacked Last Year, Data Was Stolen But the FBI Didn’t Report it

Morrissey Technology – The artificial intelligence (AI) company behind ChatGPT, OpenAI, was reportedly the victim of a hack last year. Hackers at the time reportedly stole details about the company’s AI technology design. The news of this hack was revealed by The New York Times without mentioning the source of this news, and claimed “two people familiar with the incident” leaked this.

However, they claim the hackers only broke into the forum, not the core system, which drives OpenAI’s algorithms and AI framework. OpenAI reportedly disclosed the hacking incident to employees at an employee meeting in April last year. They also inform the board of directors.

However, OpenAI executives decided not to share the news with the public. They also did not report this incident to the FBI or other law enforcement. According to The New York Times, OpenAI did not disclose the hack to the public because information about customers was not stolen.

“Executives did not consider the incident a threat to national security because they believed the hacker was an individual with no ties to the foreign government,” the newspaper said.

The New York Times source said some OpenAI employees worry China-based adversaries could steal the company’s AI secrets and pose a threat to US national security.

Leopold Aschenbrenner, the leader of OpenAI’s supersignment team at the time, reportedly shared similar sentiments about weak security and being an easy target for foreign adversaries.

Aschenbrenner said he was fired earlier this year for sharing internal documents with three external researchers to get “feedback.” He insinuated his dismissal was unfair; he scanned the document for any sensitive information, adding that it was normal for OpenAI employees to contact other experts for a second opinion.

However, The New York Times points out that studies conducted by Anthropic and OpenAI reveal that AI is “not significantly more dangerous” than search engines like Google.

However, AI companies must ensure that their security is tight. Lawmakers are pushing for regulations that impose huge fines on companies whose AI technology causes social harm.

https://www.eajpnv-ordizia.org/

https://www.encom-training.com/-/for4d/

https://anugerahprodukpertanian.com/

OpenAI Chatgpt Apple
News

Apple Didn’t Pay OpenAI to Put ChatGPT in Siri

Morrissey Technology – At WWDC 2024 yesterday, Apple announced a collaboration with OpenAI to integrate ChatGPT in Siri. Apparently there is no contract worth millions of dollars behind this collaboration. According to a Bloomberg report, Apple did not pay OpenAI to embed ChatGPT in Siri, and vice versa. Apple believes the exposure OpenAI receives by placing its technology on hundreds of millions of devices will be far greater than paying in cash.

Apple will use OpenAI’s GPT-4o model to power AI features in iOS 18, iPadOS 18, and MacOS Sequoia. Thanks to this integration, voice assistant Siri can become smarter at understanding more complex user questions, such as creating menu ideas, summarizing articles, or finding photos based on descriptions.

ChatGPT on Apple devices can also be used to write stories, as well as rewrite and proofread existing text. These features can be accessed for free without having to log in to ChatGPT, but there are several additional benefits if users subscribe to ChatGPT Plus. So, here OpenAI can earn money if Apple users subscribe to ChatGPT Plus at a cost of USD 20 per month. If users sign up for ChatGPT Plus on Apple devices, the iPhone maker will also receive a commission.

In the future, Apple hopes to receive more revenue from AI through profit sharing agreements with other companies. Apple plans to take a portion of the revenue its partners receive from monetizing the output generated by chatbots in its ecosystem. Apple FOR4D believes the number of users switching to AI from search engines will increase. This means they could get less revenue from their billion-dollar contract with Google.

Apart from OpenAI, Apple is also in discussions with Google to embed Gemini in iOS this year. The company created by Steve Jobs also approached the AI ​​company Antrophic to provide chatbot Claude as another option. A Bloomberg report claims Apple is considering collaboration with Baidu and Alibaba to master chatbot features in China, where ChatGPT and other Western-made chatbots are not officially available.

ChatGPT
News

What is the Difference between GPT-3.5 and GPT-4 in ChatGPT

Morrissey Technology – ChatGPT is one of the most popular artificial intelligence (AI) in the world. To support its users, ChatGPT has released its newest AI language, namely GPT-4, in March 2023. Previously, ChatGPT used an AI language called GPT-3.5 which can be accessed for free. However, detikers who want to use GPT-4 must subscribe first. So, what is the difference between GPT-3.5 and GPT-4? Is it true that the newest AI language from ChatGPT is more feature-rich and powerful? Check out the explanation in this article.

Differences between GPT-3.5 and GPT-4

Even though both are language model programs in conversation or dialogue format, it turns out there are differences between GPT-3.5 and GPT-4. Quoting the Coursera page, here are a number of differences:

1. Can use images

In GPT-3.5, users must write prompts using text to generate AI output from ChatGPT. Meanwhile, in the GPT-4 version, you can use text and images. So, you can enter commands in the form of text and images. After that, GPT-4 will provide an explanation or answer to the text or image you sent as a prompt FOR4D. For example, you include a picture of the inside of the refrigerator and its contents. Later, AI will suggest what food or drink recipes can be made with the ingredients in the image.

2. Produce safer and more accurate answers

OpenAI claims that GPT-4 is more secure in generating answers written to prompts than the GPT-3.5 version.

“82 percent less likely to respond to requests for unauthorized content and 40 percent more likely to produce factual responses compared to GPT-3.5 in our internal evaluation,” wrote OpenAI on its official website.

The features contained in GPT-4 are certainly good news. Because, in several cases there is often misinformation, inaccuracy and bias in the answers produced.

3. Have a bigger memory

GPT-4 is claimed to be able to remember more conversations with its users than GPT-3.5. Previously, GPT-3.5 had limitations in conveying information that was only available before 2021 and had a limited short-term memory, namely around 8,000 words. For comparison, GPT-4 is claimed to have eight times more memory than the GPT-3.5 version, which can store 64,000 words. Then, GPT-4 also has a longer context window or amount of text that can be processed simultaneously. In fact, the GPT-4 Turbo version can process up to 128,000 words. This feature can help users perform various tasks, such as including a website link in the prompt and asking ChatGPT to consider the source.

4. Wider General Knowledge

The general knowledge base in GPT-4 is said to be better than GPT-3.5. It allows users to generate, edit, and repeat various tasks such as composing songs, writing screenplays, or learning the user’s writing style. In addition, GPT-4 has used the latest data up to April-December 2023, so it can provide better and more appropriate answers. However, it all depends on the model version too.

5. Price

The final difference is in terms of price. Currently, GPT-3.5 is available in a free version and can be accessed by anyone. Even so, the features feel incomplete and not up to date. Meanwhile, GPT-4 can be accessed by subscribing every month. If detikers are interested, you have to spend US$ 20 or around Rp. 320 thousand per month. ChatGPT FOR4D also provides a ‘Team’ plan targeted at workers or students who need more features. Each person will be charged US$ 25 or the equivalent of Rp. 400 thousand per month.