Artificial Writing but SignedPhoto by Jimmy Liao on Pexels
AI News

Artificial Writing but Signed

Expresso3 September 2026 at 07:00

The central issue of the article is the use of texts written by artificial intelligence or other artificial means that are subsequently signed by someone who did not produce them. The author argues that accepting and signing prose that does not represent the recognizable expression of who signs it implies a renunciation of what would be the authorial mark.

The article maintains that, by accepting this type of practice, one implicitly assumes that only the content of the text matters, disregarding form as a distinctive element of an author's identity. This represents a fundamental shift in the understanding of what it means to produce and publish a text.

The author reflects on the ethical and journalistic implications of this practice, questioning the legitimacy of texts without genuine authorial expression in the context of journalistic production. The signature, in this case, loses its traditional meaning of linking author and work.

The article's conclusion is that texts produced artificially and signed by third parties should not have a place in newspapers, since they violate the fundamental principle of the relationship between who writes and what is written.

Related articles

AI News

AI: the challenge is no longer to implement, it is to create trust

Hugo Bastos, writing from the front lines of generative AI evolution, addresses the issue of trust in AI models and agents. He argues that this trust must be well calibrated, avoiding both blind faith and automatic skepticism, and that the true current challenge is not implementing AI, but generating that trust appropriately.

SAPO Tek03/09/26, 19:09
AI News

OpenAI launches GPT-6 Astra, new frontier AI that uses controversial reasoning technique

OpenAI launched GPT-6 Astra, a new frontier AI that uses a controversial reasoning technique called "recurrent depth," generating concerns among cybersecurity experts due to the opacity of machine decision-making. The model presents exceptional results on standardized tests, such as FrontierMath Tier 4 (97.6%), ARC-AGI (98.6%) and ExploitBench (100%), competing directly with Anthropic's Fable 5.1. GPT-6 Astra targets applications in scientific work, game development, financial services, data analysis, design engineering and business intelligence.

oglobo03/09/26, 19:00
AI News

Artificial Intelligence: ChatGPT, Grok and Claude Down Worldwide

Chatbots from three of the major artificial intelligence companies — ChatGPT, Grok and Claude — faced worldwide technical problems this afternoon, causing outages and service disruptions for millions of users globally.

Now03/09/26, 17:11
Meo testou uma campanha duas vezes. Numa delas, os consumidores não eram reais
AI News

Meo tested a campaign twice. In one of them, the consumers were not real

Meo tested the same advertising campaign twice, using the same stimuli and metrics, but with different participants: in one version, 1,200 real consumers participated, and in the other, 1,200 "cognitive replicas," meaning consumer profiles generated by artificial intelligence. The goal was to compare the effectiveness of the results obtained with both groups.

SAPO Notícias03/09/26, 17:04